This action might not be possible to undo. Are you sure you want to continue?

E LECTROMAGNETIC F IELD T HEORY E XERCISES

P LEASE VERY

NOTE THAT THIS IS A

PRELIMINARY DRAFT !

Draft version released 15th November 2000 at 20:39

Downloaded from http://www.plasma.uu.se/CED/Exercises

Companion volume to

**E LECTROMAGNETIC F IELD T HEORY
**

by Bo Thidé

**E LECTROMAGNETIC F IELD T HEORY Exercises
**

Please note that this is a VERY preliminary draft!

Tobia Carozzi Anders Eriksson Bengt Lundborg Bo Thidé Mattias Waldenvik

Department of Space and Plasma Physics Uppsala University and Swedish Institute of Space Physics Uppsala Division Sweden

Ipsum

Σ

Sweden All rights reserved. Electromagnetic Field Theory Exercises ISBN X-XXX-XXXXX-X . Copyright c 1998 by Bo Thidé Uppsala.A This book was typeset in LTEX 2ε on an HP9000/700 series workstation and printed on an HP LaserJet 5000GN printer.

. . . . Solution . . .C ONTENTS Preface 1 Maxwell’s Equations 1. . . .2 Invent your own gauge . . . . . . . . . . Solution . . . . . . . . . . . . . . . . . . . . . . . . . 2 Electromagnetic Potentials and Waves 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Example 2. 17 17 i Draft version released 15th November 2000 at 20:39 . . Example 1. . . . . . . . . . . . . . . . . . . . . Formulae used . . . . . . . . . . . . . . . . . . .3 The charge continuity equation . . . . . . . . . . . . . . . . . . Example 1. . . . . . Solution . . . . . . . . .3 Fourier transform of Maxwell’s equations Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solved examples . . . . . . .1 1. . . . Solved examples . . . . Solution . . . . . . .1 2. . . . . . . . . . . .3 Coverage . .1 The Aharonov-Bohm effect . . . . . . . . . . . . . . .1 Coverage . . . . . . . . . . . . . . . . .1 Macroscopic Maxwell equations . . . . . . . . . . .2 2. . . . . . . . . . . . . . .4 Simple dispersion relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 Maxwell’s equations in component form Solution . . . . . . . . . 3 Relativistic Electrodynamics 3. . . . . . . . Example 2. . . . . . . . . . . . . . . . . . . 9 9 9 9 9 10 11 11 13 13 15 15 Example 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Example 2. . . vii 1 1 1 1 1 2 4 4 5 5 Example 1. . . . . . . . . . . . . . . . . . . . . . . Formulae used .3 Coverage . . . . . Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 1. .

. . . . . 4 Lagrangian and Hamiltonian Electrodynamics 4. .1 6. . . . . . . . . . . . . . . . . Example 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Example 4. . . . . . . . . . . 17 18 18 18 Example 3. . . . . . . 27 27 27 28 28 28 29 29 5 Electromagnetic Energy. . . . . . . Solution . . . . . . . . . . . . . . . . . . . . . .1 EM quantities potpourri . . . . . . . . . . . . . . . . . . . . . .3 Coverage . . . . . . . . . . . . . . . . . . . . . .1 5. . . . . . . .3 Coverage . . .2 Invariant quantities constructed from the ﬁeld tensor 20 Solution . .3 Formulae used . . . . . . . . . . . . . . . Solved examples . . . . . . Solved examples . . .2 4. .1 4. . . . 20 Example 3. . . Solution . . . . . . . . . . . . . Solution . . . . . . . . . . . . . . . . . . Example 5. . . . . . . . . . Example 5. . 31 31 31 32 32 32 35 35 37 37 39 39 6 Radiation from Extended Sources 6. . . . . . . . . . . . . . . . . . . . . . . Solution . . . . . . . . . . . . . . . . . . . . . . .2 6. . . . . . . Example 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Formulae used . . . Solution . .3 Solar sailing . . . . . . . . . . . . . . Solved examples . . . . . . . . . . . . . . . . . . . . . . . . . . . Momentum and Stress 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solved examples . . . . Example 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 Covariant formulation of common electrodynamics formulas . . . . . . . . . . . . 21 21 23 23 Solution . . . . . . . . . . .4 Magnetic pressure on the earth . . . . . . . . . Example 3. . . . . . . . . .2 5.1 Instantaneous current in an inﬁnitely long conductor 42 Draft version released 15th November 2000 at 20:39 . . . .2 Classical electron radius .3 Coverage . . . . . . . . . . . . . . . . . . . . Solution . . . . . . . . . . . . . . . . . . . . . . . .1 Canonical quantities for a particle in an EM ﬁeld . . . . . . Example 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solution . . .2 Gauge invariance of the Lagrangian density . . . . . . . . . . . . . . .1 Covariance of Maxwell’s equations . . . . . . . . . . . . . . . . . 41 41 41 42 Example 6. . .4 Fields from uniformly moving charge via Lorentz transformation . . . . . . . Solution . . . .ii 3. . . Formulae used . . . . . . . . . . . . . . . . . . . . .2 3. . Formulae used . . . . . . . . .

. . . . . . . . . .2 7. . . . Example 7. . . . . . . . .3 Coverage . . . . . . .1 Rotating Electric Dipole . . . . . . . .3 Atomic radiation . . . . . . . . . . . . . . . . . Solution . . .3 The Larmor formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 Coverage . . . . . . . . Example 8. . . . . . . . . . . . . . . . . . Solved examples . . . . . . .iii Solution . 42 47 47 50 50 51 51 7 Multipole Radiation 7. . . . . . . . . . . . . . . . . . Example 6. . ˇ Example 8. . . . Formulae used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solution . . . . . . . . . . . . . . . . . Solved examples . . . . . . . . . . . . . . . . . Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Draft version released 15th November 2000 at 20:39 71 71 71 72 . . . . . . . . . . . . . . . . . . . . . Solution . . Solution . . . . . . . . . . . . . . . . . Solution . . . . . . . . . . . . . . . . . . . . 53 53 53 54 54 54 56 56 58 58 59 59 8 Radiation from Moving Point Charges 8. . . . . . . . . . . . .2 8. 9 Radiation from Accelerated Particles 9. . . . . . . . . . . . . . . . .1 8. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solution . . . . . . . . . . Example 6. . . . . . . . . . . . . . Solution . . . . . . . . . . . . . . . . . . . . . . . . .3 Travelling wave antenna . . . . . . . . . . . . . . . . . . . . . . . . Example 8. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solved examples . . . . . .1 Poynting vector from a charge in uniform motion 63 63 63 64 64 64 66 66 67 67 69 69 Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 Classical Positronium . Solution .4 Vavilov-Cerenkov emission . . . . . .4 Microwave link design . . . . . . . . . . . .2 Rotating multipole . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Formulae used . . . . . . . . . . . . . . . . Example 6. . . . . . . .3 Coverage . . . . . . . Example 7. . . . . . Example 7. . . . . . . Example 7. . . . . . . . . . . . . . . . . . . . . . . . . Example 8. . . . . . . . . . .2 Synchrotron radiation perpendicular to the acceleration .2 9. . . . . .1 9. . . . . . . . . . . . . . . . .1 7. . . . . . . .2 Multiple half-wave antenna . . . . . Formulae used . Solution .

. . . .1 Motion of charged particles in homogeneous static EM ﬁelds . . . Solution . 79 . . . . . . . . . . . . . . . . . . . . . . . . Example 9. . . . . . . .Example 9. . 77 . Example 9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 Solution . . . . . .2 Solution . 79 Draft version released 15th November 2000 at 20:39 iv . . 72 . 77 Radiation loss of an accelerated charged particle . Example 9. 74 Radiation and particle energy in a synchrotron . . . . . . . . . . . . .3 Solution . . . . . . . . . 72 Radiative reaction force from conservation of energy 74 . .

. . . . . . . . . . . . . . . . . . 43 44 47 74 v . . . . . . . . . Snapshots of the ﬁeld .2 6. . . . .3 The turn-on of a linear current at t 0 .1 6. . . Draft version released 15th November 2000 at 20:39 ¢ £¡ 6. . . . . . . Multiple half-wave antenna standing current . . . . . . . . . . . .L IST OF F IGURES 9. . . .1 Motion of a charge in an electric and a magnetic ﬁeld . .

Draft version released 15th November 2000 at 20:39 vi .

1999 B. T. The problems and their solutions were created by the co-authors who all have taught this course or its predecessor. Sweden December.P REFACE This is a companion volume to the book Electromagnetic Field Theory by Bo Thidé. Draft version released 15th November 2000 at 20:39 vii . Uppsala. It should be noted that this is a preliminary draft version but it is being corrected and expanded with time.

viii P REFACE Draft version released 15th November 2000 at 20:39 .

refreshing our knowledge of vector analysis in vector form and in component form.1 Coverage In this lesson we examine Maxwell’s equations.2c) 1 ∂ E c2 ∂ t (1.1 (1. 1.2 Formulae used ¢ ! " ¢ B E B 1.2b) (1.1b) (1.1d) 1 . the cornerstone of electrodynamics.2a) (1. We start by practising our math skill.L ESSON 1 Maxwell’s Equations 1.3 Solved examples E XAMPLE 1.1c) 1 ∂ E c2 ∂ t (1.1a) (1.2d) M ACROSCOPIC M AXWELL EQUATIONS The most fundamental form of Maxwell’s equations is E Draft version released 15th November 2000 at 20:39 # B µ0 j E B ρ ε0 0 ∂ B ∂t µ0 j § ¢ ¢ ¥¦¤ ¥ ¦¤ ¨ ©¤ ¨ ©¤ E ρ ε0 0 ∂ B ∂t (1.

2 L ESSON 1. different for different media.10) (1. but it may sometimes be convenient to separate the sources of the ﬁelds (the charge and current densities) into an induced part. due to “free” charges and currents not caused by the material properties.11) (1.7) (1.12) (1.13) . M AXWELL’ S E QUATIONS sometimes known as the microscopic Maxwell equations or the Maxwell-Lorentz equations. the induced sources are described by To fully describe a certain situation. one also needs constitutive relations telling how P and M depends on E and B. due to the response of the medium to the electromagnetic ﬁelds.3) (1.8) (1. One then writes The electric and magnetic properties of the material are often described by the electric polarisation P (SI unit: C/m2 ) and the magnetisation M (SI unit: A/m).2a) and (??) reduce to known as the macroscopic Maxwell equations. and an extraneous. Show that by introducing the ﬁelds the two Maxwell equations containing source terms (1.6) (1. In the presence of a medium. Solution If we insert and # # j ρ jind ρind jext ρext # H D ρext jext ! # D H ε0 E P B µ0 M # % ! $ jind ρind ∂P ∂t P ∂ D ∂t Draft version released 15th November 2000 at 20:39 # # j ρ jind ρind jext ρext (1.4) M (1. These are generally empirical relations. these equations are still true.5) (1. In terms of these.9) (1.

19) (1. The inherent difference lies in how a material is treated. or even to use atomic-scale calculated D and H although this is a rather useless procedure. S OLVED EXAMPLES 3 into we get which can be rewritten as ρext Now by introducing the D and the H ﬁelds such that the Maxwell equations become The reason these equations are known as “macroscopic” are that the material properties described by P and M generally are average quantities.18) ∂ P ∂t M 1 ∂ E c2 ∂ t (1. Draft version released 15th November 2000 at 20:39 7 E ND OF EXAMPLE 1.20) ε0 E (1.25) (1.14) (1. not considering the atomic properties of matter.26) QED . so the nomenclature “microscopic/macroscopic” is somewhat misleading.16) (1.24) (1.17) (1.1 6 D ρext # H ! H # D ε0 E P B M µ0 jext ∂ D ∂t ) # ' # 5) 43 B M µ0 ε0 E P jext ) ! ' # ! E 1 ρ ε0 ext P ∂ P ∂t # 0) # # ' B µ0 jext ( E ρ ε0 # ' ! & B ρind µ0 j # % 1 2 jind P 1 ∂ E c2 ∂ t ∂ P ∂t M (1.22) (1. there is nothing in principle preventing us from using largescale averages of E and B.3. not including details around single atoms etc. However.1.15) (1.23) (1. not in the spatial scales. Thus E and D get the character of averages.21) (1.

one should to some extent at least think in those terms. Even if one does not follow this complicated way always. ˆ ˆ Remember that though they are real numbers. ˆ ˆ ∑3 1 E j x j j where the last step assumes Einstein’s summation convention: if an index appears twice in the same term.28) (1. On the other hand. Even though all the equations contain vectors.27) (1. B.4 L ESSON 1.30) (1. we are certain to get the correct answer: the free index in the resulting equation necessarily comes out the same on both sides.e.33) (1. Vector quantities.2 M AXWELL’ S EQUATIONS IN COMPONENT FORM Express Maxwell’s equations in component form. The three E j are the components of the vector E in the coordinate system set by the three unit vectors x j . M AXWELL’ S E QUATIONS E XAMPLE 1.29) 1 ∂ E c2 ∂ t (1. only the latter pair are true vector equations in the sense that the equations themselves have several components. E. the E j are not scalars. can always be expanded as E E j x j. and are not to be summed over. it is to be summed over. but by using this careful procedure. The E j are real numbers.31) (1. Nabla operations are translated into component form as follows: Draft version released 15th November 2000 at 20:39 ) ' G xk ˆ G j x j xk ˆ ˆ G j δ jk Gk H xk ˆ Hi x i x k ˆ ˆ Hi δik Hk G H 8 # # # B ! $ E µ0 j ) % % ' E B ρ ε0 0 ∂ B ∂t (1. When going to component notation. while the x j are vectors. for example E. Solution Maxwell’s equations in vector form are written: In these equations. Indices which only appear once are known as free indices. symbol is used for a summation index is immaterial: it is always true that a i bi since both these expressions mean a1 b1 a2 b2 a3 b3 a b. Let us go into ridiculous detail in a very simple case: This is of course unnecessarily tedious algebra for an obvious result. What ak bk .34) (1. geometrical objects. unless i k or if a is the null vector. and j are vectors.32) (1. the expression ai ak is in general not true or even meaningful.35) . all scalar quantities are of course left as they are. Such an index is called a summation index. i. Vector equations are transformed into component form by forming the scalar product of both sides with the same unit vector. while ρ is a scalar.

38) ∂ Ei ρ ∂ xi ε0 ∂ Bi 0 ∂ xi ∂E εi jk k ∂xj (1.39) (1. an equation of the form Ai B j is almost invariably in error! With these things in mind we can now write Maxwell’s equations as T HE CHARGE CONTINUITY EQUATION Derive the continuity equation for charge density ρ from Maxwell’s equations using (a) vector notation and (b) component notation.40) ∂ B ∂t i 1 ∂ Ei c2 ∂ t (1. a derivation of the continuity equation for charge looks like this: Compute ∂ ∂t E ∂ ∂t in two ways: 1.3 .1. Apply to Gauss’s law: C) % ' ∂ ∂t E 1 ∂ ρ ε0 ∂ t Draft version released 15th November 2000 at 20:39 (1.42) µ 0 ji E ND OF EXAMPLE 1. Remember that in vector valued equations such as Ampère’s and Faraday’s laws.2 E XAMPLE 1. As said above.3.41) (1. Compare the usefulness of the two systems of notations. Also. discuss the physical meaning of the charge continuity equation. S OLVED EXAMPLES 5 ∂φ ∂ xi ∂ Vi ∂ xi εi jk where V is a vector ﬁeld and φ is a scalar ﬁeld.37) % ∂ Vk ∂xj (1.36) (1. Solution Vector notation In vector notation.43) 7 # 9 A! # B µ0 j 1 ∂E c2 ∂ t εi jk ∂ Bk ∂xj ! " 9 A! ! B E ∂B ∂t 9 A! B 0 9 A! E ρ ε0 9 @! % ( V 9 @! 9 @! ∇φ ∂ Φ ∂ xi ∂ V V ∂ xi i ∂ εi jk x i ˆ V ∂xj k xi ˆ (1. one must be careful to make sure that the free index on the left hand side of the equation is the same as the free index on the right hand side of the equation.

48) and that µ0 ε0 c2 1: (1.45) 1 c2 ∂ E ∂t C) % ' ' D F 0 9 @! (1.6 L ESSON 1.50) .47) 1 ∂ ∂ Ei c2 ∂ x i ∂ t ∂ ∂ xi . Take the divergence of the Ampère-Maxwell law: B Comparison shows that Component notation In component notation.49) (1. M AXWELL’ S E QUATIONS 2.46) (1. Take the divergence of the Ampère-Maxwell law: Comparison shows that Draft version released 15th November 2000 at 20:39 G # ∂ ρ ∂t ∂ ji ∂ xi ! " ∂ ∂ Ei ∂ t ∂ xi 1 ∂ j ε0 ∂ x i i 0 E Use that the relation εi jk Ai A j 0 is valid also if Ai # I H ∂ ∂ xi εi jk ∂ ∂ Ei ∂ t ∂ xi 1 ∂ ρ ε0 ∂ t ∂ Bk ∂xj G % # ∂ ρ ∂t j 0 µ0 ! " ∂ E ∂t 1 ε0 ∂ j ∂ xi i j E C) Use 0 and µ0 ε0 c2 # µ0 j 1: (1.44) (1. a derivation of the continuity equation for charge looks like this: Compute ∂ ∂ ∂ xi ∂ t Ei ∂ ∂t in two ways: 1. (1. Take of Gauss’s law: 2.

integrate the continuity equation over some volume V bounded by the surface S. or at least to our picture of it. while in the component notation. Although not seen here. In the vector notation system.51) ∂t is known as a continuity equation. for which vector notation becomes cumbersome. we ﬁnd that dQ ∂ j dS ρ d3x j d3x (1.. while with the component system you need only the deﬁnitions of εi jk and δi j . we sometimes need to keep some vector formulas in memory or to consult a math handbook. ∇φ is ∇φ in any coordinate system. the components depend on the unit vectors chosen. The vector system with div. curl etc. i. S OLVED EXAMPLES 7 Comparing the two notation systems We notice a few points in the derivations above: Interpreting the continuity equation The equation ∂ ρ j 0 (1. the continuity equation is the ﬁeld theory formulation of the physical law of charge conservation.e.1. E ND OF EXAMPLE 1.3.52) dt ∂t V V S which says that the change in the total charge in the volume is due to the net inﬂow of electric current through the boundary surface S. may be closer to the physics. Draft version released 15th November 2000 at 20:39 7 Q ! T" Q ! S" Q R # P P P P It is sometimes difﬁcult to see what one is calculating in the component system. Why? Well. the component system of notation is more explicit (read unambiguous) when dealing with tensors of higher rank. Hence.3 . By using Gauss’s theorem. The vector notation system is independent of coordinate system.

Draft version released 15th November 2000 at 20:39 8 .

the Maxwell equations can be reduced to algebraic equations via the Fourier transform and the physics is contained in so-called dispersion relations which set the kinematic restrictions on the ﬁelds.2) 9 .1 Coverage Here we study the vector and scalar potentials A and φ and the concept of gauge transformation.1) (2.3 T HE A HARONOV-B OHM Consider the magnetic ﬁeld given in cylindrical coordinates. Seen as wave equations. One of the most important physical manifestation of Maxwell’s equations is the EM wave.1 r0 θ z r0 θ z Bˆ z 0 (2.L ESSON 2 Electromagnetic Potentials and Waves 2. Draft version released 15th November 2000 at 20:39 X) W W X) W W V ' Y ' Br Br ¢ B ¨ U¤ ¢ Solved examples EFFECT E XAMPLE 2.2 E Formulae used A ∇φ ∂A ∂t 2. 2.

This problem naturally divides into two parts: the part within the magnetic ﬁeld and the part outside the magnetic ﬁeld. Also. let us use the gauge φ 0.6) r ∂r Integrating this equation we ﬁnd that Moving to the outer problem.8) . Before we discuss this effect let us calculate the vector ﬁeld from the given magnetic ﬁeld.3) (2. from which we can derive the EM ﬁelds. The equations connecting the potentials with the ﬁelds are In this problem we see that we have no boundary conditions for the potentials. Examining the The ﬁrst equation tells us that A is time independent so A other three we ﬁnd that there is no dependence on θ or z either so A A r .7) (2. Could it be that the potentials and not Maxwell’s equations are more fundamental? Although the ultimate answer to these questions is somewhat metaphysical.5a) (2. we see that the only difference compared with the inner problem is that B 0 so that we must consider ) 1 ∂ rAθ r ∂r ' Aθ Br 2 0 Draft version released 15th November 2000 at 20:39 ) ' ) W W ' 3 ! ) ) ' ' 1 1 r ∂A ∂t ∂ Aθ 1 ∂ Az r ∂θ ∂z ∂ Ar ∂ Az ∂z ∂r ∂ rAθ ∂ Ar ∂r ∂θ ! ! B A ! E φ ∂A ∂t ( ! $ (2. E LECTROMAGNETIC P OTENTIALS AND WAVES Determine the vector potential A that “generated” this magnetic ﬁeld. it is exactly these questions that make the Aharonov-Bohm effect.5c) (2. All that remains is 1 ∂ rAθ B (2.4) 0 0 0 B (2. Solution A interesting question in electrodynamics is whether the EM potentials φ and A are more than mathematical tools. and alternatives to the Maxwell equations.5b) (2. Let us start with the interior part: A r θ z .10 L ESSON 2.5d) (2.

Then invent your own gauge and verify that it is indeed a gauge condition.7) the arbitrary constant C and can write in outer solution as Now in electrodynamics (read: in this course) the only measurable quantities are the ﬁelds.12) (2.9) r If we demand continuity for the function Aθ over all space we ﬁnd by comparing with (2.3. S OLVED EXAMPLES 11 This time integration leads to C (2. The interpretation is that the potential is a more fundamental quantity than the ﬁeld. even though there is no magnetic ﬁeld along their path.11) ∇φ ∂ A ∂t (2. the Aharonov-Bohm effect shows that this situation does have a measurable consequence.1 E XAMPLE 2. So the situation above.13) µε ∂ φ ∂t µσ φ µj (2. In quantum mechanics however.14) 7 ` Aθ Aθ 2 Br0 2r 0! (2. I NVENT YOUR OWN GAUGE Name some common gauge conditions and discuss the usefulness of each one. Solution Background The Maxwell equations that do not contain source terms can be “solved” by using the vector potential A and the scalar potential φ . where we have a region in which the magnetic ﬁeld is zero but the potential is non-zero has no measurable consequence in classical electrodynamics. when letting charged particles go around this magnetic ﬁeld (the particles are do not enter the magnetic ﬁeld region because of a impenetrable wall) the energy spectrum of the particles after passing the cylinder will have changed. and j σ E j (where j is the free current density arising from D other sources than conductivity) and deﬁnitions of the scalar and vector potentials in the remaining two Maxwell equations and ﬁnd that Draft version released 15th November 2000 at 20:39 a ! "b3 # # ! ! ! ∇2 A µσ µε ∇ @1 ∂ A ∂t ! " # ∇2 φ ∂A ∂t ρ ε ∂ 2A ∂ t2 A a a # ! ! & E B A (2.10) E ND OF EXAMPLE 2. we can use the constitutive relations εE H B µ . Namely.2 . isotropic and homogeneous media.2. deﬁned through the relations Assuming linear.

e. In this case (2. The deﬁnitions of the scalar and vector potentials are not enough to make A and φ unique. since the deﬁnitions (2.11) and (2. but if one is given B and E there many ways of choosing A and φ . This kind of transformation is called a gauge transformation and the fact that gauge transformations do not affect the physically observable ﬁelds is known as gauge invariance. i.17) (2.12) determine B and E. And once we have found A and φ it is straight forward to derive the E and B ﬁelds from (2.19) (2.20) (2. but the B and E ﬁelds do not change. Furthermore.16) A µσ φ φ 0 0 0 A 0 then (2. This can be seen through the fact that A and φ can be transformed according to where ψ is an arbitrary scalar ﬁeld.11) and (2. Gauge conditions The ambiguity in the deﬁnitions of A and φ can be used to introduce a gauge condition. if one is given A and φ then (2.15) (2. Some common gauge conditions are A The Lorentz gauge is the most commonly used gauge for time-varying ﬁelds. for static ﬁelds.14) reduce to So the Lorentz transform decouples (2.13) and (2.14.14) reduce to ! $ 3 ! ! ∇2 ! $ 3 ! ! ∇2 ! & ∇2 A ! & ∇2 φ ρ ε µj ∂ ∂t ∂ µσ ∂t µσ ∂2 φ ∂ t2 ∂2 µε 2 A ∂t µε ρ ε µj 0. .12) do not completely deﬁne A and φ we are free to add certain conditions.12).13) and (2. Using (2.13) and (2.21) .13) and (2. E LECTROMAGNETIC P OTENTIALS AND WAVES These equations are used to determine A and φ from the source terms.14) and puts φ and A on equal footing.11) and (2. Draft version released 15th November 2000 at 20:39 T( # 1 ∂ 2A c2 ∂ t 2 A µj In the temporal gauge one “discards” the scalar potential by setting φ (2.12 L ESSON 2.18) 1 1 (2. reduces to % # µε∂ φ ∂ t # Coulomb gauge Lorentz gauge Temporal gauge % ! a φ φ # a A A ∇ψ ∂ ψ ∂t (2. In other words. the resulting equations are manifestly covariant. In this gauge The Coulomb gauge is most useful when dealing with static ﬁelds.

3 4 c a a E ND OF EXAMPLE 2. Furthermore. Of course. To perform this derivation we need formulas on how to translate the operators and ∂ ∂ t in Maxwell’s equations. is in many case more fundamentally described as a function of angular frequency ω than length x.e.23) i k x ωt F t x (2. which satisfy your condition.2 . So given a A and a φ which satisfy the physical conditions through (2. The Fourier transform in time. Once you have an equation that you think might be a gauge. S OLVED EXAMPLES 13 Thus the single vector A describes both E and B in the temporal gauge. By transforming them we get simple algebraic equations instead of differential equations. Lorentz and the temporal mentioned above are rarely used in introductory literature in Electrodynamics.3.2.24) 7 E XAMPLE 2. the Fourier transformed Maxwell equations are useful when working with waves or time-varying ﬁelds. To verify that a condition is a gauge condition it is sufﬁcient to show that any given set of A and φ can be made to satisfy your condition.13) and (2. F OURIER TRANSFORM OF M AXWELL’ S EQUATIONS Fourier transform Maxwell’s Equation. such waves are called static waves. Solution Maxwell’s equations contain only linear operators in time and space. is deﬁned analogously and so a combined spatial and time Fourier transform becomes ∞ Draft version released 15th November 2000 at 20:39 ) W ' h e Q d) W ' ˜ F ω k dt d3x e e f g e ∞ ) ' e Q C) ' E ˜ Ak d3x e ik x ) ' f e e Q d) ' E f˜ ω ∞ dt eiω t f t (2. is deﬁned by ∞ ∞ ∞ .14) we try to see if it is possible (at least in principle) to ﬁnd a gauge transformation to some new potential A and φ . a gauge condition is at least a scalar equation containing at least one of the components of A or φ . it must be veriﬁed that it is a gauge. the dielectric function. Use the Fourier version of Maxwell’s equations to investigate the possibility of waves that do not propagate energy. This is done through gauge transformations. This makes it easy to Fourier transform them. but it is instructive to consider what constitutes a gauge condition by constructing ones own gauge. i. especially since the response function. How to invent your own gauge Gauges other than Coulomb. and the Fourier transform in space.22) Ax (2.

We then get. For 0 ˜ iεi jk k j Ek ω k ei where we have once again used partial integration. One may proceed analogously for ∂ ∂ tE t x . (2. These transformation rules are easy to remember by saying that roughly the Fourier transform of ∇ is ik and the Fourier transform of ∂ ∂ t is iω .27a) (2.28a) (2. H and B ﬁelds.27c) on Maxwell’s equations.27a). Trivially.14 L ESSON 2. after some simple trimming Draft version released 15th November 2000 at 20:39 ) W ' X) ) W ' $X) ! ! ) W ' X) ) W ' X) W ' W ' W ' W ' k ik k k ˜ E ˜ H ˜ D ˜ B ω ω ω ω k k k k ˜ ωB ω k ˜ω k ˜ j iω D ω k ˜ iρ ω k 0 (2.27b) (2. one gets similar equations for the transformation of the D.28d) ! dt Ei e h dt d3x iεi jk k j Ek t x e ) W ' e f g e x Q S! h i k x ωt ! 9 ) W ' 8 v wu t GG wG e f g e h dt d3x e i k x ωt εi jk ∂Ej t x ei ∂ xk i k x ωt ) W ' ) W ' ) W ' ! ¦9 W ' ) ) W ' 9 5) W ' ) W ' 9 5) W ' % e f g e Q Q Q 9 E dt d3x e ( h i k x ωt Et ω x s i x ∞ E we get ) y) W ' !' Q ! dt xe ( q rh ) W ' C) W ' ) W ' Q 8 v wu t ) W ' p Q e f g e ∑ Ei t i k x ωt h dt d3x e ) W ' e f g e e f g e e Q e i Q 9 E ∞ ∞ dt d3x e i k x ωt h i k x ωt E we get E e f g e ) W ' ) W ' % ∂ Ei t x ∂ xi dt d3x iki Ei t x (2.28c) (2.25) ei (2.26) .27c) where V t x is an arbitrary ﬁeld and denotes here “Fourier transform”. E LECTROMAGNETIC P OTENTIALS AND WAVES If we apply the last formula on ∞ ∞ 0 iki dt d3x Ei t x ˜ iki Ei ω k ˜ ik E ω k where we have used partial integration.27b) and (2.28b) (2. Thus we have found that Vt x Vt x ∂V t x ∂t ˜ ik V ω k ˜ ik V ω k ˜ iω V ω k (2. Now we can use (2.

As an example of the use of the Fourier transformed Maxwell’s equations let us derive static waves. usually something like From this one can solve for ω which is then a function of k1 k2 k3 .33) 7 ) W W ' C) W ' Bα 0 ωα Bα kα Eα 0 E ND OF EXAMPLE ` ` (2. homogenous.30) 2. we see that S E H 0 trivially. From (2.e.4 .29) (2. what combinations of k and ω ) are possible. what is the dispersion relation in a conducting medium? Solution A dispersion relation is a relation between ω and k. what is the dispersion relation? And what is the group velocity in this case? Also.3. S OLVED EXAMPLES 15 where we have dropped the ˜ notation. S IMPLE DISPERSION RELATION If a progressive wave is travelling in a linear.3 ( E XAMPLE 2. homogeneous medium with σ 0 (so j σ E 0).28a) this implies that So. then these equations give Draft version released 15th November 2000 at 20:39 # H ! & E ∂B ∂t ∂D j ∂t (2. The lesson here is that you can have time-varying ﬁelds that do not transmit energy! These waves are also called longitudinal waves for obvious reasons.32) (2. The dispersion relation is derivable in principal once one has explicit knowledge of the dielectricity function (or response function) for the medium in question. These are the Fourier versions of Maxwell’s equation. nonconducting dielectric medium with dielectricity parameter ε and permeability µ . Static waves are one possible oscillation mode for the E and H ﬁelds. isotropic. so kα Eα . Let’s say that we have a mode α such that the E Eα ﬁeld is oscillating at ω ωα 0 and that it has a k kα 0 which is parallel to the electric ﬁeld. For isotropic media then ω will be a function of k k only. isotropic. A dispersion relation determines what modes (i.2. The two vector equations in Maxwell’s equations are Dω k 0 (2.31) so for a progressive wave characterized by ω and k propagating in a linear.

! 4 ∂ω ∂k 2u2 k ! iσ 1 4u2 k2 2ε 2 and the group velocity is ω σ2 ε2 ! # ω 2 σ i ω ε uk 2 2 ! ! " k2 E iσ µω E ! ! $ k k E B µ ωB iσ E ωε E ω 2ε µ E E 0 where we have identiﬁed the phase velocity u 1 C) 4 ' 4 ω ω2 ! k2 ω 2ε µ 0 k2 εµ k ku εµ εµ.36) (2.34) (2.44) (2.38) F F F F (2. the two vector equations applied on a wave which at ﬁrst resembles the progressive wave we used above gives Combining these two equation as done previously. we get So that Draft version released 15th November 2000 at 20:39 7 If σ 4u2 k2 σ 2 ε 2 0 then we are back again to the previous problem as can be veriﬁed.37) (2. ) ! k2 ω 2ε µ E 0 ! " ! C) ) x vwu ' t ' k k E k E ωk B k k2 E ! $ k k F ! ' 8 E E B µ ωB ωε E and then using (2.16 L ESSON 2.40) (2.43) (2.47) (2.48) E ND OF EXAMPLE 2.35) ω 2ε µ E (2. E LECTROMAGNETIC P OTENTIALS AND WAVES Operating on (2.34) with k 0 Progressive wave! Since E is not assumed to be zero then The group velocity in this case is ∂ω ∂ vg uk u ∂k ∂k so in this simple case the group velocity is the same as the phase velocity.4 .41) (2.45) (2.39) (2. For the case of a conducting medium.35) we get a single vector equation: (2.46) (2.42) (2. in which j σ E.

Also. 3. show how 4-tensors are manipulated. 1.1 Coverage We examine the covariant formulation of electrodynamics.L ESSON 3 Relativistic Electrodynamics 3.1) A Lorentz boost in the 3-direction µ Lµ The ﬁeld tensor (components are 0.2 Formulae used e (3. We take up the concept of 4-tensors and give examples of these. 2.2) 17 ¢ . 3) Draft version released 15th November 2000 at 20:39 g ¢ F µν e 0 E1 E2 E3 E1 0 cB3 cB2 E2 cB3 0 cB1 E3 cB2 cB1 0 e fd g γ 0 0 γβ 0 1 0 0 0 0 1 0 γβ 0 0 γ e fd (3. We discuss the group of Lorentz transformations in the context of electrodynamics.

Solution The d’Alembert operator 1 ∂2 ∇2 (3. so Draft version released 15th November 2000 at 20:39 a a # a a # a a # a a a ∂ ∂ x3 ∂t ∂ ∂ x3 ∂ t ∂ ∂ x3 ∂ x1 ∂ ∂ x3 ∂ x1 ∂ x2 ∂ ∂ x3 ∂ x2 ∂ x3 ∂ ∂ x3 ∂ x3 a a # a a # a a # a a a ∂ ∂ x2 ∂t ∂ ∂ x2 ∂ t ∂ ∂ x2 ∂ x1 ∂ ∂ x1 ∂ x1 ∂ x2 ∂ ∂ x2 ∂ x2 ∂ x3 ∂ ∂ x2 ∂ x3 a a # a a # a a # a a a ∂ ∂ x1 ∂t ∂ ∂ x1 ∂ t ∂ ∂ x1 ∂ x1 ∂ ∂ x1 ∂ x1 ∂ x2 ∂ ∂ x1 ∂ x2 ∂ x3 ∂ ∂ x1 ∂ x3 a ) a ' ! a a ma a x1 x2 x3 t x1 x2 x3 ! E ijj jjl k h 2 vt (3. even though closest to our intuitive picture of the fabric of space-time.18 L ESSON 3.1 Solved examples C OVARIANCE OF M AXWELL’ S EQUATIONS Discuss the covariance of Maxwell’s equations by showing that the wave equation for electromagnetic ﬁelds is invariant with respect to Lorentz transformations but not Galilean transformations. we evaluate ∂ ∂ x µ ∂ xν ∂ xµ ∂ ∂ xν . The dynamics of EM ﬁelds is completely described using the d’Alembertian. does not leave the d’Alembertian invariant. Galilean transformations. i. A Galilean transformation is simply t where the origin of the primed system is moving relative the unprimed along the 3-direction with velocity v.3) c2 ∂ t 2 is a fundamental operator in electrodynamics. Now we introduce this transformation by expanding each differential in the unprimed coordinate system in terms of the differential in the primed system by using the chain rule of derivation.10) .6) (3.e.8) (3.4) (3.3 E XAMPLE 3.5) (3.9) (3. R ELATIVISTIC E LECTRODYNAMICS 3.7) (3.

18) (3. Since γ and β depend on v The 4-gradient.3.3. Thus we have found that Which obviously does not have the same form as the d’Alembertian in the unprimed system! Let us do the same calculations for the case of a Lorentz transformation. ∂µ ∂ ∂ xµ transforms as so In other words we have found that the d’Alembertian is invariant under Lorentz boosts.1 . Draft version released 15th November 2000 at 20:39 7 E ND OF EXAMPLE n n n n ) ! ' ! ! ! ) ! γ2 2 γ 2 β 2 ∂0 2 ∂1 n n n n ' 2 ∂µ ∂ µ gµν Lµ Lν ∂µ ∂ν µ ν 2 ∂2 γ2 2 γ 2 β 2 ∂3 gµν ∂µ ∂ν ∂µ ∂ µ n n n I n # H n n n ∂µ xν n ∂ xµ ∂ ∂ xµ µ ∂ µ Lν x ν ∂ µ ∂ xµ u ! n oqqr p n Lµ µ ! γ 0 0 γβ 0 1 0 0 0 0 1 0 γβ 0 0 γ tt ws n n xµ Lµ xµ µ µ ∂ Lν ∂ xµ µ Lν a a # a ! a ! a 9 2 2 ∇ 2Ψ 1 ∂ 2Ψ c2 ∂ t 2 v2 ∂ 2 Ψ c2 ∂ x32 2v ∂ 2 Ψ c2 ∂ t ∂ x 3 0 ∂µ Lµ ∂µ µ a a a ! a a # a C) a ! a ') a ! a h a ' ∂2 ∂ t2 ∂ ∂t v ∂ ∂ x3 ∂ ∂t v ∂ ∂ x3 ∂2 ∂t 2 v2 ∂2 ∂ x32 a a # a a # a a a # a ! a a ∂ ∂t ∂t ∂ ∂ x1 ∂ ∂t ∂t ∂ t ∂ x1 ∂ ∂ v ∂t ∂ x3 ∂ x2 ∂ ∂ t ∂ x2 n n n ∂2 ∂ x2 1 ∂2 ∂ x12 . 2. 1. ∂2 ∂ x2 3 ∂2 ∂ x32 (3.14) (3.15) (3.11) ∂ x3 ∂ ∂ t ∂ x3 (3.12) (3.16) (3.13) h h E 2v ∂2 ∂ t ∂ x3 (3. 3). S OLVED EXAMPLES 19 and so but and so where we have used the fact that the operators ∂ ∂ t and ∂ ∂ x3 commute.19) 3. more speciﬁcally we consider a boost along the 3 axis which is given by where (remember that µ runs over 0.17) (3. ∂2 ∂ x2 2 ∂2 ∂ x22 .

20) (3. R ELATIVISTIC E LECTRODYNAMICS E XAMPLE 3.24) cB and B E c. What can we create from µν and F µν which is invariant under Lorentz transformations? We consider the obviously invariant quantities µν µν Draft version released 15th November 2000 at 20:39 ! w9 9 We see that the dual tensor can be obtained from F µν by putting E v u ! ! v ! ! ! ! 0 cB1 cB2 cB3 cB1 0 E3 E2 cB2 E1 0 E1 cB3 E2 E1 0 u ! tt s ! o pqqr Fµν gµα gνβ F ! ! oqqr ! p αβ 0 E1 E2 E3 E1 0 cB3 cB2 E2 cB3 0 cB1 E3 cB2 cB1 0 tt ws u ! ! ! 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 u ! tt s ! ! o pqqr F µν ! ! ! 0 E1 E2 E3 E1 0 cB3 cB2 E2 cB3 0 cB1 E3 cB2 cB1 0 tt s o pqqr αβ 1 αβ γδ ε Fγδ 2 1 αβ γδ ε gγ µ gδ ν F µν 2 v v v P P (3.22) (3.23) (3. can a progressive wave be seen as a purely electric or a purely magnetic ﬁeld in an inertial system? Solution The dual tensor of Fαβ is given by where and gµν We determine ﬁrst the ﬁeld tensor with two covariant indices through the formula so αβ From the formula for the dual tensor we see that it is a 4-tensor since F is a four tensor and ε is easily shown to be an invariant under orthogonal transforms for which the Lorentz transform is a special case.2 I NVARIANT QUANTITIES CONSTRUCTED FROM THE FIELD TENSOR 1 αβ γδ Fγδ of the ﬁeld tensor Fµν . are invariant under Lorentz transformations? Having found these quantities you should be able to answer the questions: can a purely electric ﬁeld in one inertial system be seen as a purely magnetic ﬁeld in another? and.21) (3.20 L ESSON 3. . What quantities Construct the dual tensor αβ 2ε constructed solely with the ﬁeld tensor and it’s dual tensor.

E XAMPLE 3.25) (3. For a progressive wave E B so X 0 and in a purely electric or a purely magnetic ﬁeld 0 X 0 also. E c p. C OVARIANT FORMULATION OF COMMON ELECTRODYNAMICS FORMULAS Put the following well know formulas into a manifestly covariant form The continuity equation Lorentz force The inhomogeneous Maxwell equations The homogenous Maxwell equations Solution The Methodology To construct manifestly covariant formulas we have at our disposal the following “building blocks”: Also we have the 4-gradient Draft version released 15th November 2000 at 20:39 ) W ) W ) W ) W ) W W ) W ) ' ' ' ' ' ' ' an event 4-velocity 4-momentum wave 4-vector 4-current density 4-potential 4-force xµ uµ pµ kµ Jµ Aµ Fµ ct x .3. ρc j . All inertial systems must have the same value for X and Y .26) c2 B2 are Lorentz invariant scalars.3.3 . In other words it does not seem that a purely electric ﬁeld can be a purely magnetic ﬁeld in any inertial system. γc γv .2 ya ! E E ! This means that E B and E 2 ) ! { x ' ! " Fµν 4cE B E2 2 c 2 B2 v z ma P P P P ! v µν (3. Let us say that X E B and Y E 2 c2 B2 . but for a progressive wave E cB so Y 0 and if the other system has E or B 0 then Y 0 force both the ﬁelds to be zero. So this is not possible. F µν Fµν Relation of EM ﬁelds in different inertial systems Now that we know that E B and E 2 c2 B2 are Lorentz invariant scalars. let see what they say about EM ﬁelds in different inertial systems. ω c k. A purely magnetic ﬁeld would mean that E 0. so X 0 but Y 0. γv F c γF 7 E ND OF EXAMPLE ma 3. so X 0 and Y 0. φ c A. A purely electric ﬁeld in one inertial system means that B 0. S OLVED EXAMPLES 21 µν Fµν and F µν Fµν .

to accomplish our goal.22 L ESSON 3. To get a vector quantity from F µν and J µ we contract these so our guess is Inhomogeneous Maxwell equations The inhomogeneous Maxwell may be written as the 4-divergence of the ﬁeld tensor ∂α F αβ Fµ ∂µ J µ 0 F µν Jν Jβ Draft version released 15th November 2000 at 20:39 u ! ! o pqqr ﬁeld tensor ! F µν ! ! ! 0 E1 E2 E3 E1 0 cB3 cB2 E2 cB3 0 cB1 E3 cB2 cB1 0 tt s ) (3. thus in space-time the continuity equation is simply stated as the 4-current density is divergence-free! Lorentz force We know that the left hand side of Lorentz force equation is a 3-force. 4 and the 4-component is the time-like component. Obviously we should use the covariant 4-vector force instead.27) (3.e. we will need to use our knowledge of the formulas we will try to make covariant.29) W ' E 4-gradient ∂µ ∂ ∂ xµ ∂ ∂ ct ∂ ∂ x .e. And on the right hand side of th Lorentz equation is a 3-vector quantity involving charge density and current density and the E and B ﬁelds. i. there is also the system where indices run 1. One can always raise or lower a index by including a metric tensor g αβ . The EM ﬁelds are of course contained in the ﬁeld tensor F µν .28) (3. Beware! A sufﬁcient condition to formulate covariant electrodynamic formulas is that we make our formulas by combine the above 4-vectors. 1. On top of this sufﬁcient condition. This leads us to calculate the contraction of the outer product between the 4-gradient ∂µ and the 4-current J ν This is covariant version of the continuity equation. R ELATIVISTIC E LECTRODYNAMICS as our operator building block and also the second rank 4-tensor Observe that we use indices which run 0. In the notation we use here contractions must be between a contravariant (upper) index and a covariant (lower) index. 3 where the 0-component is time-like component. The continuity equation We know that the continuity equation is a differential equation which includes the charge density and the current density and that it is a scalar equation. 2. equate two indices and sum over this index (notationally this means we create a repeated index). 3. To make sure we have a covariant form we take outer product (i. simply combine the tensors so that all the indices are free) and then perform zero or more contractions. 2.

S OLVED EXAMPLES 23 Homogeneous Maxwell equations The homogenous Maxwell equations are written most compactly using the dual tensor of the ﬁeld tensor.32) (3. Solution We wish to transform the EM ﬁelds. The EM ﬁelds in a covariant formulation of electrodynamics is given by the electromagnetic ﬁeld tensor where we are using components running as 1. 2. rectilinear motion q 1 v 2 c2 R0 E 4πε0 s3 are obtained via a Lorentz transformation of the corresponding ﬁelds in the rest system of the charge. 4.3.3 .31) (3.35) 7 E XAMPLE 3.4 tt s ! ! v ∂α αβ 0 E ND (3.30) OF EXAMPLE 3.3.34) (3. Using the dual tensor we have F IELDS FROM UNIFORMLY MOVING CHARGE VIA L ORENTZ TRANSFORMATION In the relativistic formulation of classical electrodynamics the E and B ﬁeld vector form the antisymmetic electrodynamic ﬁeld tensor 0 Ex Ey Ez Ex 0 cBz cBy Ey cBz 0 cBx Ez cBy cBx 0 Show that the ﬁelds from a charge q in uniform. 3. To transform the EM ﬁelds is to transform the ﬁeld tensor. A Lorentz transformation of the ﬁeld tensor can be written where Draft version released 15th November 2000 at 20:39 u ! o pqqr Lν µ ! γ 0 0 βγ 0 1 0 0 0 0 1 0 βγ 0 0 γ tt s F ~F0 L L Fλσ Lλ Lσ F ν µ µ ν u ! ! ! ! oqqr F ! µν ! 0 cB3 cB2 E1 cB3 0 cB1 E2 cB2 cB1 0 E3 E1 E2 E3 0 tt s B v E c2 u ! ) ! ' ! ! ! pqqr o ! Fµν ∂ µ Aν ∂ν A µ (3.33) (3.

45) (3.43) (3.46) (3. R ELATIVISTIC E LECTRODYNAMICS so that F µν 0 0 0 0 E1 0 0 0 0 E2 0 0 0 0 E3 0 E1 0 E2 0 E3 0 A little matrix algebra gives 0 0 0 0 E1 0 1 0 0 0 0 0 0 E2 where s vt Draft version released 15th November 2000 at 20:39 ) # γ ct βx u ' oqqr p u oqqr L y z ) ! ' u x0 y0 z0 ct 0 x y z ct γ x s tt ws tt ws tt ws ! $ B3 B2 B1 E1 E2 E3 0 E1 0 γ E2 0 γ E3 0 γβ 0 E c 3 γβ 0 E c 2 u ) # ! ' ! ! ! 0 0 β γ 2 E1 β γ 2 E1 0 γβ E2 0 γβ E3 2 0 γ 1 β 2 E1 0 γβ E2 0 0 0 γ E2 0 γβ E3 0 0 0 γ E3 0 β 2 E1 0 γ E2 0 γ E3 2 0 0 β γ E1 β γ 2 E1 γ2 1 u tt s ! ! ! ! ! orq ! q u ! γ 0 0 βγ 0 0 1 0 βγ 0 0 γ 0 β γ E1 0 β γ E2 0 β γ E3 0 γ E1 0 0 0 0 E2 0 0 0 0 E3 0 γ E1 0 γ E2 0 γ E3 0 β γ E1 tt s u ! qoq r u tt s ! ) # ! ! ' ! ! ! ! ! qoq r o wqqr o wqqr ! ~FL L ~ L ! 0 0 0 0 E3 0 E1 0 E2 0 E3 0 γ 0 0 βγ 0 1 0 0 0 0 1 0 βγ 0 0 γ tt s tt s tt ws u ! ! ! oqqr p B0 0 # # E0 # q x0 x y 0 y ˆ ˆ 4πε0 x0 2 y0 2 z0 z ˆ z0 2 3 # ! oqqr where γ 1 1 β 2 and β v c where v v1 x.37) (3.41) (3.47) (3.36) (3.48) .39) (3.40) (3.44) (3. The ﬁelds in the rest system S0 are ˆ (3.42) (3.38) (3.24 L ESSON 3.

S OLVED EXAMPLES 25 E R2 0 q γ x s x y y zˆ ˆ ˆ z 4πε0 γ 2 x s 2 y2 z2 3 s x 2 y2 z2 γ2 γ2 Draft version released 15th November 2000 at 20:39 7 ! R3 0 ) ! ' E q 1 4πε0 β2 R0 1 β 2 sin2 θ 3 ) ! ' R2 1 0 β 2 sin2 θ ~) ! ') # ' # R2 0 y2 z2 1 γ2 1 # ! ! # ) ! ' x s 2 R2 0 y2 z2 # # y2 z2 # ) ! ' x s 2 y 2 z2 γ2 } | E q 4πε0 γ 2 C) ! ' ! s x R0 sin θ π R0 cos θ 2 R0 3 ) # # # # ) ! ' # ) ! ' ' ! y' # ) (3.3.49) (3.55) (3.4 OF EXAMPLE .3.54) (3.51) (3.50) (3.53) (3.52) y2 z2 (3.56) E ND 3.

Draft version released 15th November 2000 at 20:39 26 .

We look at both the point Lagrangian for charges in EM ﬁelds and the Lagrangian density of the EM ﬁelds.2 Formulae used qφ The Lagrangian for a charged particle in EM ﬁelds is A useful Lagrangian density for EM ﬁeld and its interaction with charged particles is given by Draft version released 15th November 2000 at 20:39 E2 A j ¥ 1 ε c2 B2 2 0 ¢ ¢ L qv A ¥ mc2 γ (4.L ESSON 4 Lagrangian and Hamiltonian Electrodynamics 4.1) ρφ (4. 4.1 Coverage We brieﬂy touch the Lagrangian formulation of electrodynamics.2) 27 .

1 Solved examples C ANONICAL QUANTITIES FOR A PARTICLE IN AN EM FIELD Derive the canonical momentum and the generalised force for the case of a charged particle mc2 in EM ﬁeld given by φ and A.6) 3 # ! ! # ! # 1 ! & ! & ! & ! " Pi ∂L ∂ vi ∂ ∂ vi mc2 1 vi vi c2 qφ qv j A j # ! ! " # ! ! " F F (4.9) .4) L γ we ﬁnd that mγ vi qAi P p qA qv A. γ Solution We know from analytical mechanics that the canonical momentum P is found through ∂L Pi (4. L AGRANGIAN AND H AMILTONIAN E LECTRODYNAMICS 4.28 L ESSON 4. which gives us.8) A (4. # qE qv B Draft version released 15th November 2000 at 20:39 # # q φ qv A q ∂A ∂t ) ' ! ! ) ' # ) S ' # q φ qv A qv A q qv f (T ! ) f ' # ! & Q q φ q v A ! # q ∂φ ∂ xi q ∂ v jA j ∂ xi q dAi dt dA q dt ! # ! ! & Qi ∂ qφ ∂ xi qv j A j d ∂ qφ dt ∂ vi qv j A j ∂A ∂t ! And on the other hand the generalised force is ∂U d ∂U Qi ∂ xi dt ∂ xi ˙ where U is the generalised velocity dependent potential qφ ! c2 1 vjvj c2 # # | mc2 vi qAi (4.3) ∂ vi so with mc2 qφ qv A (4.3 E XAMPLE 4.7) (4. The Lagrangian is L qφ qv A.5) (4.

Remembering that E and B are invariant under gauge transformations we ﬁnd that E2 A ' 0 Draft version released 15th November 2000 at 20:39 ! ) ψj ∂ ρψ ∂t (4.14) ∂ ρψ ∂t ∂ρ ψ (4. Is it invariant? If not. discuss the consequences this would have.11) (4. but what is it? G AUGE INVARIANCE OF THE L AGRANGIAN DENSITY Consider the Lagrangian density of the EM ﬁelds in the form We know that EM ﬁeld are invariant under gauge transformations Determine the Lagrangian density under a gauge transformation. This charge will perform gyro-harmonic motion. Obviously the momentum is not conserved but on the other hand we did not expect it to be conserved since there is a force on the charge.4.13) (4.2 . So we expect P p qA to be a constant of the motion. which is true in the case we have here since the potentials do not depend on spatial coordinates. However we now from analytical mechanics.16) ∂t 7 # ' # ! B E ND OF EXAMPLE 4.17) ! ) ! ) E2 A j ρφ ψ ∂ρ ∂t j ψj # ! ) E2 ! ) ! ! ! 1 ε c 2 B2 2 0 1 ε c 2 B2 2 0 1 ε c 2 B2 2 0 E2 A j A j ' ' ! ) # # a # a % % ' !) #£) ' ! a # a % ' ! a ! ) # a # a ' ) ! ρφ ρφ ψ j ρ ∂ψ ∂t ψj ψ j ) ! a ' # ) ψ ! ' % ! " a ' ' ' ! ! a ! ! ! 9 φ φ # a A A ψ ∂ψ ∂t 1 ε c 2 B2 2 0 # ! ) E2 A j ! 1 ε c 2 B2 2 0 ρφ (4. S OLVED EXAMPLES 29 What does the canonical momentum P p qA represent physically? Consider a charge moving in a static magnetic ﬁeld.10) (4.3.15) ∂t ∂ ρψ (4. that it is the conservation of canonical momentum that is more general.1 E XAMPLE 4. Conservation of canonical momentum is found when the problem is translational invariant. Solution Let us insert the gauge transformation relations into the Lagrangian density.12) j ρ φ ∂ψ ∂t (4.

Furthermore. but what about the other two terms? Some thought reveals that the neccessary condition for a Lagrangian to be physically acceptable is not the Lagrangian itself is invariant but rather that the variation of the action integral S d3x dt is invariant. So now we would like to check that the gauge transformations indeed do not affect any the variation of the action. Let us use the ﬁrst alternative. L AGRANGIAN AND H AMILTONIAN E LECTRODYNAMICS where we have used the continuity equation. gauge transformations of the potentials do not effect the physics of the problem! Draft version released 15th November 2000 at 20:39 7 δ Sext 0 Q Sext ρψ d3x t1 t0 Q # ψ j dS dt ρψ d3x t0 t1 t0 3 # Q Q ! Q %# ψ j dS dt ρψ d3x ψ 1 3 # 0) ' Q Q Sext ψj ∂ ρψ ∂t d3x dt t1 ∂ρ ∂t 0 obviously has the same form %@1 Q Q Q Q p it is sufﬁcient to examine j d3x dt (4.2 . or two.17) into Euler-Lagrange equations to check the resulting equations differ from the Maxwell equations. we realize that we must hold t0 and t1 the end point of the particle path ﬁxed. if we assume no ﬂux source/sink at inﬁnity then we can write Now when taking the variation. Now it is possible to proceed in two different ways to do this: one is simply carry out the integration in the deﬁnition of the action integral and check that its variation is zero. one could plug in the Lagrangian density (4.30 L ESSON 4.18) (4. Since the action is linear in where we have used the continuity equation. and thus As one would expect. remembering that the variation of the action is equivalent to the Euler-Lagrange equations.19) (4.20) E ND OF EXAMPLE 4. We see that as .

1) Hi B j 1 δ ED 2 ij k k Hk Bk (5.L ESSON 5 Electromagnetic Energy.2) 31 . Momentum and Stress 5.2 ¢ S Formulae used E Poynting’s vector Maxwell’s stress tensor Draft version released 15th November 2000 at 20:39 ¢ Ti j Ei D j ¨ H (5. 5. energy.1 Coverage Here we study the force. momentum and stress in an electromagnetic ﬁeld.

4) .3) (5. k2 k3 0. intensity and stress associated with the electromagnetic ﬁelds for the following cases: Solution Background (a) Progressive wave This case is an example of a progressive or propagating wave. Assume that D c) Br Bθ µ0 m cos θ 2π r 3 µ0 m sin θ 4π r 3 b) same as in a) but for k1 iα ε E and B a) for k1 E H h E 0 e 2 e i k 1 x1 k ˆ E µω k ωt ε µω . The following table gives the relevant quantities.3 E XAMPLE 5.1. µ H. M OMENTUM AND S TRESS Table 5.32 L ESSON 5. momentum density. use the real part in the formulas. If the given ﬁelds are complex. 1 2E 5. Since E and H are complex we must ﬁrst take their real parts: The energy density is Draft version released 15th November 2000 at 20:39 ) ! ' ! " Re H k1 E cos k1 x1 µω 0 ) ! ' Re E E0 cos k1 x1 ω t e2 ω t e3 Also.1 Solved examples QUANTITIES POTPOURRI EM Determine the instantaneous values of the energy density. The ﬁeld vectors in this table are assumed to be real. E LECTROMAGNETIC E NERGY. ) # ' ! # Ei D j # Name Energy density Intensity Momentum density Stress Symbol Uv S PEM T Formula D 1H B 2 E H εµE H Hi B j 1 δi j Ek Dk 2 Hk Bk SI unit J/m3 W/m2 kg/s m2 Pa e g (5. identify these cases.

15) (5.11) (5.7) (5.5) (5.9) (5.3.8) (5. We take real part The energy density is Draft version released 15th November 2000 at 20:39 ! ) "C ' e ! $ Re H E0 e e ! $ Re E E0 e α x1 sin ω t e2 α e Im E µω 1 α E 0 α x1 e cos ω t e3 µω of the ﬁelds keeping in mind the fact that k1 ! " e ! T33 T32 iα is imaginary: α e µω 1 α x1 ) ! ' ! ) ! ' ! T22 T21 ) ! ' ! ) ) ! ! ' ' ! ! B T11 ε 2 µ 2 E H 2 2 2 3 2 ε 2 k1 2 E cos2 k1 x1 ω t E0 cos2 k1 x1 ω t 2 µω 2 0 2 ε E0 cos2 k1 x1 ω t T31 T12 0 ε 2 µ 2 E H 2 2 2 3 2 ε 2 k1 2 E0 cos2 k1 x1 ω t E cos2 k1 x1 ω t 2 µω 2 0 T13 T23 0 µ 2 ε 2 H E T22 0 2 3 2 2 ) ! ' ! B ! B PEM εµS 2 ε ε µ E0 cos2 ω ε µ x1 ωt ω t e2 ) ) ! ! ε 2 E cos2 ω ε µ x1 µ 0 ε 2 E cos2 ω ε µ x1 µ 0 e3 cos ω t e2 C) ) ! ' !' ) ) ' ' ! ' ' S E0 cos k1 x1 ω t e2 k1 E cos k1 x1 µω 0 ω t e3 ~) ! ' ) # £) ! ! ' ' ! $ ! $ Uv 2 ε 2 k1 E 2 cos2 k1 x1 E0 cos2 k1 x1 ω t 2 2µω 2 0 2 ε E0 cos2 ω ε µ x1 ω t ωt (5.13) (5.6) (5.14) (5.5.12) (5.17) . S OLVED EXAMPLES 33 The intensity or power density is ω t e1 The momentum is The stress is (b) Evanescent wave This case is a example of an evanescent wave.10) (5.16) (5.

The stress tensor 3 # ) # C) # ' # ' Uv 1 1 B2 2 µ0 r 1 2 µ0 2 µ0 m2 4 cos2 θ 16π 2 r6 2 µ0 m2 sin2 θ 16π 2 r6 3 ! ! " ! " ! T33 T22 1 e µ 2 H 2 3 ε 2 E 2 2 T32 2 E 0 e 2α x 1 ε sin2 ω t α cos2 ω t µω 2 2 3 ! ! T22 1 e ε 2 µ 2 E H 2 2 2 3 T13 T23 0 T21 2 E 0 e 2α x 1 ε sin2 ω t α cos2 ω t µω 2 2 3 # 1 2 E0 e e ε 2 µ 2 E H 2 2 2 3 T31 T12 0 ! " ! ! & PEM e ! $ S εµS e 2 α E0 e µω 2α x 1 sin ω t cos ω te1 2 α E0 ε e ω 2α x 1 sin ω t cos ω te1 2α x 1 ε sin2 ω t α2 cos2 ω t µω 2 3 # # 1 e Uv 2 ε 2 2α x 1 2 α 2 E 0 2α x 1 e sin ω t cos2 ω t E0 e 2 2µω 2 2 E 0 e 2α x 1 α2 ε sin2 ω t cos2 ω t 2 µω 2 e e (5.23) (5. The ﬁelds are real and in spherical coordinates.34 L ESSON 5.18) (5.20) (5.21) (5. M OMENTUM AND S TRESS The intensity is The momentum is The stress is T11 2 2 2 (c) Magnetic dipole This case is a magnetic dipole.19) (5.22) (5.25) (5.26) (5. The energy density is 2 Bθ µ0 m 2 1 4 cos2 θ sin2 θ 6 32π 2 r µ0 m 2 1 3 cos2 θ 32π 2 r6 Draft version released 15th November 2000 at 20:39 The intensity is S components are 0 since E 0.24) (5. E LECTROMAGNETIC E NERGY.27) . and likewise for the momentum.

30) (5.2 .36) E ND OF EXAMPLE µ0 m 2 8 cos2 θ 1 3 cos2 θ 32π 2 r6 µ0 m 2 1 5 cos2 θ 32π 2 r6 µ0 m2 sin θ cos θ 1 Br Bθ µ0 8π r6 0 Tθ r ) 5. From relativity one can show that fundamental particles should be point-like. We then relate this ﬁeld energy to the mass of the electron.32) (5. S OLVED EXAMPLES 35 µ0 m2 cos2 θ 4π 2 r6 µ0 m 2 1 32π 2 r6 C LASSICAL ELECTRON RADIUS Calculate the classical radius of the electron by assuming that the mass of an electron is the mass of it’s electric ﬁeld and that the electron is a homogenous spherical charge e.29) (5.31) 3 cos2 θ (5. This points to the fact that Maxwell’s equations has a minimum length scale validity where quantum mechanics takes over.33) (5.5. if one calculates the electromagnetic mass from Maxwell’s equation one gets an inﬁnite result due to the singularity in Gauss law E ρ ε0 . Solution One the “failures” of Maxwell’s equations or classical electrodynamics is on question of mass of point particles. We will calculate this problem as follows: determine the electric ﬁeld in all of space and then integrate the formula for the energy density of the electric ﬁeld.35) (5. This integral will contain the radius of the electron since it partitions the integration.1 # ' ! ) ! ! ! ' ' ! ! & Trr 1 2 B µ0 r Uv 3 cos2 θ E XAMPLE 5.3. which is a known quantity. We use this relation to solve Draft version released 15th November 2000 at 20:39 7 Trφ Tθ φ Tφ φ Tφ θ ) µ0 m 2 2 sin2 θ 1 3 cos2 θ 32π 2 r6 µ0 m 2 1 3 cos2 θ 2 sin2 θ 32π 2 r6 0 0 0 0 ) # ' ) ! # ! ! ! ' ' ! ! & Tθ θ Tφ r Trθ 1 2 B µ0 θ Uv Tθ r µ0 m2 sin2 θ 16π 2 r6 µ0 m 2 1 32π 2 r6 ) (5. However.28) (5. Mass and energy are related through the distribution of radius re and total charge q equation E mc2 .34) (5.

That is all space 4π 0 re 0 0 0 re Finally.37) 4πε0 r2 and the region inside the sphere is given by Gauss law which in this case gives If we integrate this equation. Introduce spherical coordinates.41) (5.37).44) .39) and furthermore it is continuous with (5. We may now determine the energy density of the electric ﬁeld due to the electron.40) (5.e. is given by Coulomb’s law: e E r re (5.39) (5. r re . It is simply Uv r re Now we integrate Uv over all space.43) (5. re 3 e2 5 4πε0 me c2 Draft version released 15th November 2000 at 20:39 U m e c2 3e2 1 20πε0 re me c2 (5. M OMENTUM AND S TRESS for the “electron radius”. (5.42) re d3x ∞ re e2 1 2 r dr dΩ 32π 2 ε0 r4 (5. We start by determining the electric ﬁeld. Consider a homogeneous spherical charge distribution in a vacuum. We divide the whole space into two parts: the region outside of the sphere. we relate the total electric ﬁeld energy to the rest mass of the electron and solve for the electron radius. we ﬁnd that e r Er r r e 3 4π re ε0 We can easily verify that (5.39).45) ! # e2 r5 6 8πε0 re 5 re e2 8πε0 1 r ∞ 3e2 1 20πε0 re Q Q # e2 r2 r2 dr dΩ 6 32π 2 ε0 re 4π ) Y ' # ) V ' Q Q U Uv r C) C) V ' Y ' ©F Q Uv 1 ε E2 2 0 Uv r 1 ∂ 2 r Er r2 ∂ r e 3 4 3π r e ε 0 re C) C) V ' Er re ρ ε0 1 e 32π 2 ε0 r4 2 e 2 6 r 32π 2 ε0 re 2 re Uv r Y C) V ' Y ' % F F (5.38) (5.36 L ESSON 5. thus making the solution unique. E LECTROMAGNETIC E NERGY.e.40) is indeed a solution to (5. i. i.

Draft version released 15th November 2000 at 20:39 Now the continuity equation for EM momentum says that ∂ PEM ∂ t the component along the sail surface normal this implies that ! " pafter ^ EM n pbefore ^ n EM ! " pbefore n ˆ EM S ∆A ∆t cos2 θ c S ∆A ∆t cos2 θ c ) ' C) ' pbefore EM Pbefore ∆V EM S c2 ∆A c ∆t cos θ (5. If we assume the sail to be a perfect reﬂector. The momentum carried by such a wave is so the momentum along the direction of the surface normal n is ˆ After hitting the sail.2 G E XAMPLE 5. Let us characterize this wave by a Poynting vector S of duration ∆t. It travels in space with velocity c.49) ∂ Pmech F so for 7 & # ! re e2 4πε0 me c2 (5. If the continuity equation for electromagnetic momentum ∂ PEM ∂ t T ∂ Pmech ∂ t is integrated over all space the stress term disappears and what is left says that a change in electromagnetic momentum is balanced by mechanical force. the pulse will be characterized with all the same quantities as before the impact.1 mm thick aluminium coating as material for a sail. S makes an angle 180 θ with the normal of the sail surface. except that the component along the surface normal will be opposite in sign. At 1 AU (= astronomical unit distance between sun and earth) the intensity of the electromagnetic radiation from the sun is of the order 1 4 103 W/m2 . Imagine a localised pulse of a progressive electromagnetic wave incident on a plane. S OLAR SAILING Investigate the feasibility of sailing in our solar system using the solar wind. S OLVED EXAMPLES 37 This last result can be compared with the de facto classical electron radius which is deﬁned as and is found by calculating the scattering cross section of the electron. furthermore. That the wave is progressive means simply that it is “purely radiation” (more about progressive waves in the next lesson). it “lights up” the area ∆A on the surface on the sail. 2 mm in thickness. with a 0. Thus. what would the acceleration be for different incidence angles of the sun’s EM radiation on the sail? Solution In this problem we will use the principle of momentum conservation.48) (5. It would weigh 1 g/m2 . One technical proposal uses kapton.47) (5.46) E ND OF EXAMPLE 5. and.5.3 . This is because the sail is assumed to be perfectly reﬂecting.3.

38 L ESSON 5. At 1 AU.3 kW m2 .52) 3 mm/s2 (5.3 So ﬁnally. the solar constant is 1.50) e . M OMENTUM AND S TRESS pbefore S EM 2 n ^ 2 c ∆A cos θ ∆t The other force components are zero by symmetry. Draft version released 15th November 2000 at 20:39 7 e a 0 9 10 5 N/m2 3 10 3 kg/m2 e 2 13 3 03 W/m2 cos2 θ 108 m/s (5. E LECTROMAGNETIC E NERGY. the pressure exerted by the pulse on the sail is simply S cos2 θ c (5. weighs 3 10 3 kg/m2 ) due to the solar wind can at most be which is half of the acceleration due to the sun’s gravitational ﬁeld. From equation (5.51) G G F ^ n ! pafter EM (5. The solar constant is the intensity of EM radiation or in other words. The solar wind can be seen as a multitude of such pulses radiating radially outwards from the sun. we ﬁnd that Newton’s second law gives the acceleration a of the sail (which. as we recall. the magnitude of the Poynting vector S .51). n F ^ ∆A 2 We now have the pressure exerted by a pulse charcterized by S incident on a surface with an angle θ .53) E ND OF EXAMPLE 5.

which for this case is So by taking the inner product of T and the unit vector pointing in the direction the north pole. according to the hypothesis in problem. Solving for T1 we arrive at ª $ © so that 1 P0 P1 0 1 ∝m∝B∝ Tperiod G 6 1 4 10 e e G B2 r 2 µ0 6 10 5 2 2 1 3 10 1 3 10 6 G ¤¥ £¦! ¡ e G ' ) e ¤¥ ¨ § ¡£ ¢ T B 2 µ0 r 0 0 0 0 0 0 0 0 e G © F E XAMPLE 5. S OLVED EXAMPLES 39 M AGNETIC PRESSURE ON THE EARTH Determine the magnetic pressure due to the Earth’s magnetic ﬁeld at the magnetic poles (take B 6 10 5 T) and compare this with the Earth’s atmospheric pressure (1 atm 1 01 105 Pa).57) and T0 (5.3. we ﬁnd that 3 ª T1 T0 0 1 10 s E ND OF EXAMPLE where we denote the current values of the pressure and rotational period time. The static magnetic pressure is quantiﬁed in the Maxwellian stress tensor.55) (5.54) 3 Pa (5. we ﬁnd the pressure in the radial direction to be Now we assume. respectively and the hypothetical values 1 and T1 . but also through static ﬁelds. that 0 1 Draft version released 15th November 2000 at 20:39 7 T1 24 3600 1 4 10 105 e G So with 105 Pa.58) (5. how much faster would the angular velocity need to be for the magnetic pressure to be comparable to the atmospheric pressure? Solution In this exercise we see that EMF can exert pressure not only via radiation pressure.59) 5.5.4 .4 Uv 0 0 0 Uv 0 0 0 Uv (5.56) (5. Now assume that the magnetic dipole moment is proportional to the angular velocity of the earth. namely 1 0 0 .

Draft version released 15th November 2000 at 20:39 40 .

2 Formulae used At x d3x General expressions for retarded potentials For “antennas”. one-dimensional current distributions.L ESSON 6 Radiation from Extended Sources 6. And for monochromatic. which we will call antennas here. This problem is solved in different ways depending on the explicit form of the source distribution. we use Draft version released 15th November 2000 at 20:39 ± ²® ¡ ® ¢ Eω ¯ ³ ° ¯ ik sin θ eik x x 4πε0 c x x ± ²® ¡ ® ¢ Bω ¯ ° ¯ iµ0 k sin θ eik x x 4π x x ˆ ϕ ˆ θ ® 41 ¬ ¢ « φ t x 1 4πε0 d3x ¡ ® ¡ ¡ «¡ ρ x t x x ® ¡ ® ¡ « ¡ ¡ jx t x x ¬ µ0 4π ¢ « . we use the general expressions for the retarded potentials.1 Coverage We will study the important general problem of how to calculate the EM ﬁelds induced by spatially extended. time-varying sources. we use the formulas given below. For truly extended bodies with non-monochromatic time dependence. 6.

Solution This problem belongs to the most general category of problems of the kind where given a source distribution one wants to ﬁnd the EM ﬁelds. E and S (6.3 Solved examples E XAMPLE 6. conducting wire 0. which in the Lorentz gauge take the form Q C) W ' φ t x d3x Draft version released 15th November 2000 at 20:39 a ! a )a Wa ' 1 4πε0 Q d) W ' At x d3x a ! a ) ra W a ' µ0 4π jt x x x ρ t x x x It is assumed that the conductor can be kept uncharged.1 I NSTANTANEOUS CURRENT IN AN INFINITELY LONG CONDUCTOR Consider the following idealised situation with an inﬁnitely long. This is because the source is not monochromatic. it is current free.42 k2 sin2 θ µ0 c 32π 2 x x 2 L ESSON 6. ρ in whole space.2) . Determine B.1) (6. R ADIATION FROM E XTENDED S OURCES where l l 6. V x W W ¶C) ' ¡ jω z e ikz cos θ ¡ ° ¡ V ® ° ¬ ¢ µ ¢ S ® ± ® ® ´ ± 2 r ˆ dz 0 . so it is not an antenna.e. and furthermore it is an extended distribution. For t applied simultaneously over the entire length of the wire. thin.. so multipole expansion analysis is not possible. So we must use the most general formula for calculating ﬁelds induced by time-varying sources. the wire carries the current 0 t 0 j z t 0 J Hint: Calculate ﬁrst the vector potential A . Consequently. but at time t 0 a constant current J is along the z axis. i.

which is illustrated in Fig.3. For the remaining z integration we use cylindrical coordinates (see Figure BLP1. The “switching on” at t 0 can be written as a step function θ t . the expressions for the sources consist of a time-dependent part times a space-dependent part.6. 6. .1). So we have for the current that and for the charge density (6. the space dependent part can be written δ x δ y J z.4) x c. This can be seen as and charges ﬂowing in opposite directions such as to keep the total charge density ρ t x 0 but this could still have a total current density j t x 0. And if we orient the wire along the z axis.1. The function t x x c is known as the retarded time tret .1) is to ﬁnd the explicit form for j t x and ρ t x .1. We use cylindrical coordinates. ˆ On the other hand. the charge distribution ρ t x 0 as given in the problem formulation. perform the integrations to obtain the potentials A t x and φ t x and. S OLVED EXAMPLES 43 where the source time t is to be replaced by t t x x c so t is seen as a function of t.3) (6. The current density distribution j is along the z axis and is turned on at t 0. x and x . In many problems. ﬁnally.5) z r Figure 6. derive the E and B ﬁelds from the potentials in the normal fashion and the Poynting vector from the ﬁelds. Let us ﬁnd the explicit expression for the current density j x t . This is one such case. The integrations over x and y are trivial: µJ ∞ θ t x zz c ˆ z 0 ˆ At x dz 4π x zz ˆ ∞ a a a a ! ) a ! ! ' a a e Q a C) a W a ' ρ t x 0 x )a ' )a ' )a ' C) a W a ' jt x δ x δ y Jθ t z ˆ ` W ' ) ) W ' Aa a a ) ra W a ' ¸·a ! D! ) W ' C) W ' ya ) ra ' ra ' ) ) ¹a ' a ! ! ) ra W a ' ! C) W ' a # ) ra W a ' C) W ' a ) W ' (6. remembering to replace the source time t with t and perform the integration.2cyl) so we can write Draft version released 15th November 2000 at 20:39 ºa ! ! We insert (6. The algorithm of solution now that we have decided to use (6.3) into (6. an expression which is meaningful only relative the ﬁeld point t x .

which is only meaningful relative the ﬁeld point. These limits can be understood as follows (cf.2): after the current is switched on. if we assume as or where we have introduced a simply as a shorthand.12) P 2 2 r2 2 (6..11) (6.6) (6. This series of snapshots shows what the part of the current is seen at the ﬁeld point P at different ﬁeld times t. or in other words in a straight line.e.10) (6. This illustrates the concept of retarded time. 2 P r Figure 6. EM ﬁelds are sent from each point along the wire and travel at the speed of light. Since the information (i.13) .8) (6. each ﬁeld point only sees those parts of the current which are close enough.2. only those z contribute which satisfy which can be written. the current turnon) carried by the ﬁelds travel in the “line of sight”. Fig.7) F F F ! ¿ ¿ (6. This means that when integrating over z . R ADIATION FROM E XTENDED S OURCES The step function in the integrand is zero when its argument is less than zero.9) (6. So we have now that µJ a ˆ A z 0 4π a z2 r2 Draft version released 15th November 2000 at 20:39 ÏÐ Î 1 1 Î Í Î ËÌ µJ z 0 ln ˆ 4π 1 1 r2 c2 t 2 r2 c2 t 2 É 1 dz ½ ½ Í 4Ê Ê É È Ç Æ ½ Å Å ¼ t r c t t Ã Á À Ä¿ Â "¼ ¾ À ¿ 4¼ 2 r2 c E ! V a V ! c2 t 2 r2 z ! z "V a V # a z2 W Y t 0 r2 c2 t 2 r2 c2 t 2 c2 t 2 r2 a P a # Y F Y » a ! b! t rr ˆ zz c ˆ 0 ct r2 a G a z2 a # a ! x x rr ˆ x r2 z2 (6.44 L ESSON 6. and also the information gathering sphere. 6.

since ρ t x 0.21) (6. (6.3.19) (6. S OLVED EXAMPLES 45 Now that we have the vector potential we can derive the E and B ﬁelds. The B ﬁeld is and since A only has its z -component different from zero in the cylindrical system.18) F (6.6. which satisﬁes (6. the scalar potential we may set so φ t x 0.22) . Introducing f ! & ! " B A 1 r2 c2t 2 .14) (6.17) (6.15) B ϕ ∂r we then can write The electric ﬁeld is derived from But and so 2 and All we have left is to determine the Poynting vector S Draft version released 15th November 2000 at 20:39 9 9 Notice that as t ∞ then E ! | ! " E z ˆ µ0 J 2π t 1 1 r2 c2 t 2 0! ) ! ' # 3 RÑ3 ln ! ! # 1 1 ∂ ∂t 1 1 f f ∂f ∂t 1 2r2 2 f c2 t 3 1 r2 f c2 t 3 1 1 f 2 ∂t f 1 f ∂f 3 3 ln ! # 1 1 ! & E ∂A ∂t µJ ∂ z 0 ˆ 4π ∂ t 1 1 f f 2 ft ) ' Observe that for large t we have the “static” case: B ! 2π r | B ˆ ϕ µ0 J 1 r2 c2 t 2 3 Ñ3 ! # 1 1 ! $ ∂ Az ∂r ! " ∂f ∂r c2 t 2 r 1 f µ0 J ∂ ln 4π ∂ r µ0 J 2 4π r f 1 1 f f ˆ ϕ µ 0 J 2π r ) ' ! E Thus we only need ∂ Az ∂r .20) (6.16) ) Cra W a ' C) W ' On the other hand. we ˆ have ∂A ˆ z (6.

Let us see what happens if we apply the advanced potential to this problem.28) (6. The only thing that changes from the outset is that the source time t is replaced by the advanced time ta t x x c instead of the retarded time tr t x x c.29) (6. So in practice it is impossible to produce this physical setup.24) (6. This is due to the quick “turn on”.27) says we have no information about what happened after turn on. But what is interesting is that now we see that the relation (6. In the physical world only gradual turn ons are possible.1 . Physics seems to be conspiring on us in such a way that we cannot compare the advanced and the retarded potential at the same time! Draft version released 15th November 2000 at 20:39 7 a V ! W ! ! 4V a z c2 t 2 r2 ! Y bY a # a z2 z W V t 0 r2 c2 t 2 c2 t 2 r 2 c2 t 2 r2 z a ! "V Y a ! 4# t rr ˆ zz c ˆ 0 t rr ˆ zz c ˆ ºa ! $# E !ÓF a ) ! ' ·a Ò P S dC µ0 J2 2π t 1 cr22 2t ! r2 c2 t 2 2π t 1 r2 c2 t 2 ) E ND ' | ! | ) ! ' ) ' ! ! S B z ˆ 1 E µ0 µ0 J ˆ ϕ (6.30) OF EXAMPLE 6.27) (6. R ADIATION 1 µ0 J µ0 2π r 1 FROM E XTENDED S OURCES µ0 J2 r ˆ 2π 2 rt 1 cr22 2t From this expression we can for example calculate the radiated power per unit length by integrating over a cylindrical surface C enclosing the wire: We see that an inﬁnite power is transmitted starting at t 0 and r 0 which travels out to inﬁnite r.46 L ESSON 6.26) (6.9) seems to say that we have no information about what happened before turn on. In the above we tacitly employed the retarded potential without discussing the possibility of using the advanced potential. This many be seen as a consequence of what is known as Gibb’s phenomenom.25) F F a E (6. while the relation (6. With ta as the argument to the step function.23) (6. the contribution to the integral comes only from those z which satisfy If we assume that then we may write this as or One can proceed further and calculate the resulting integral.

we may use the “antenna formulae. since we have a monochromatic source and the current is an extended one-dimensional distribution. Depending on whether the length of the wire is an even (as seen in a) or odd (as seen in b) multiple of half the wave length.31) (6.32) . S OLVED EXAMPLES 47 M ULTIPLE HALF .2 (6. which is a multiple m of λ 2.6. Determine the angular distribution of the radiated electromagnetic power from the antenna. These facts give us a relation between the integration limits and k.WAVE ANTENNA half wavelengths is assumed to have a current disA wire antenna with a length of m tribution in the form of a standing wave with current nodes at the its endpoints. for the case of m even. We perform both integrations over z from l to l which is the total length L. Hint: It can me convenient to treat even m and odd m separately. As shown in Figure 6. we will use sin kz and for m odd we use cos kz . the current distribution is sin kz or cos kz .33) 2 k Draft version released 15th November 2000 at 20:39 ) ¹a ' F ) ¹a ' ! ) ¹a ' a G ) ¹a ' F F a) even sin kz b) odd cos kz a a l l z l l z ) ¹a ' ! )a ' a jω z e ) ¹a ' n ) ' e a l ikz cos θ 2 Ù e R(Ù Q CØ S k2 sin2 θ µ0 c 32π 2 r2 2 r ˆ dz Õ Ô ) Ö»' ! × E XAMPLE 6. namely πm L 2l mλ 2 l (6.” but since only the radiated effect is ask for all we need is where l so we seek a form for the current distribution jω z . It simpliﬁes matters if we consider the cases for m an even and m an odd multiple of half wavelengths separately. Figure 6. Thus.3. Solution One realises that the setup in this problem is an antenna.3.3.

34) u kz 1 k du × . R ADIATION FROM E XTENDED S OURCES Let us consider the odd case.32) gives us 2 J0 µ0 c cos2 mπ 2 cos θ r ˆ 8π 2 r2 sin2 θ Let us consider the case when m is even so the current distribution is written We remember the relation (6.36) J0 sin kz (6. So we have that Ga C) a ' jω z Ø S (6.33) which is still valid but now m is an even number.37) Draft version released 15th November 2000 at 20:39 ß) ' ) ' ) Þ X ) # ' # £) # ' # 0) ! ) ' ) ) ) ! ' ! ' ' ' ' Ý J0 k sin2 θ Ü Ü e ) h ! ' e g e ! ) J0 eiu 1 cos θ 2k i 1 cos θ e iu 1 cos θ i 1 cos θ h # eiu e iu e e e ! ' e g Ü e Q J0 2k mπ 2 iu cos θ du mπ 2 Ûa dz a J0 cos kz e dz a n e a l a ikz cos θ Ü e 2ÚÙ Q C) a ' # jω z J0 cos kz (6.35) k sin2 θ Inserting this into (6.48 L ESSON 6. The current distribution is written and so l mπ 2 mπ 2 mπ mπ sin 1 cos θ sin 1 cos θ 2 2 π π cos θ sin 1 cos θ sin 1 cos θ 2 2 J0 2 sin mπ 2 cos mπ 2 cos θ 2 cos θ cos mπ 2 sin mπ 2 cos θ k sin2 θ 2J0 cos mπ 2 cos θ (6.

39) (6.38) Þ ) # ' ) ) ! ' ) ! ' Ý # ' 1 1 1 1 ! ' ! # Ý J0 ik sin2 θ J0 ik sin2 θ cos θ sin mπ cos θ 2 cos θ sin cos θ Ü e # ' ! ' i1 i1 h ! h ! h ! h J0 ei 1 2ik e i1 cos θ ei 1 e i1 cos θ mπ 2 Ü ) } g e } g ) e g e cos θ mπ 2 cos θ mπ 2 h ! h eiu 1 } g e e g mπ 2 cos θ mπ 2 iu 1 cos θ ! e eiu e e e Ý Ü Ü e g Ü e Q e Q J0 2ik J0 2ik mπ 2 iu cos θ iu du du cos θ mπ 2 cos θ mπ 2 a a Ü à Q J0 e sin kz dz a a n e mπ 2k ikz cos θ u dz kz 1 u du ! & × ÚÙ F mπ 2 mπ 2 .6.2 Þ C # 2J0 mπ mπ cos sin cos θ 2 2 2 ik sin θ 2J0 mπ cos θ i sin 2 2 k sin θ Þ ~ ) # ' ) ! ' 1 sin 1 cos θ sin cos θ mπ 2 mπ 2 mπ mπ cos θ sin cos cos θ 2 2 cos θ ) # # ' ) sin 1 sin 1 ! cos θ mπ 2 cos θ mπ 2 (6.40) E ND OF EXAMPLE 6.3. S OLVED EXAMPLES 49 mπ 2k mπ 2 e Draft version released 15th November 2000 at 20:39 7 Ø S 2 J0 µ0 c sin2 mπ cos θ 2 r ˆ 8π 2 r2 sin2 θ 2Ù áF 2 2 4J0 mπ sin2 cos θ 2 sin4 θ 2 k (6.

Determine the angular distribution of the electromagnetic radiation from this antenna. If we insert this distribution into (6.41). Solution We need the formula k2 sin2 θ 32π 2 r2 L 2 where dz e L 2 L 2 2 Finally we have Draft version released 15th November 2000 at 20:39 7 ) ! ' Ø S ) ! ' 2 J0 sin2 θ 8π 2 r2 µ0 sin2 kL 1 cos θ 2 ε0 1 cos θ 2 r ˆ 3) ! ' ) ! ' 2 Ù 2 sin2 1 2 4J0 k2 1 cos θ kL 1 2 cos θ 3 ¸) ! ' 2 J0 sin k 1 cos θ kL 1 2 cos θ L 2 ! 1 Ü e ) ) ! ' J0 e e Ü eik 1 cos θ z ik 1 cos θ n h a ! ' e g Ü e Q J0 L 2 n h dz eik 1 e g L 2 cos θ z ) ~¹a ' In this case the distribution is given: j z we get sin α n )a ' n ik cos θ z 2 Ù e a Ü Ü e Q R(Ù Ü Ø S µ0 ε0 2 r ˆ (6. This means that the current distribution comprises travelling waves emanating from the feed point so one can assume that j z J exp ikz along the wire. The termination is adjusted such that no current is reﬂected back on the wire. 1 iα e 2i iα (6.41) j z (6. R ADIATION FROM E XTENDED S OURCES E XAMPLE 6.3 ) ra ' ) ~ra ' â ÚÙ × × .43) (6.44) (6.45) E ND OF EXAMPLE 6.42) J0 eikz .50 L ESSON 6.3 T RAVELLING WAVE ANTENNA A wire antenna of length is fed at one of its endpoints by a transmitter signal and is at its other end terminated with a resistance to ground.

6.3. S OLVED

EXAMPLES

51

Microwave links are based on direct waves, i.e., propagation along the line of sight between the transmitter and receiver antennas. Reﬂections from the ground or a water surface may in unfavourable cases distort the transmission. Study this phenomenon using the following simple model: The transmitter antenna T is a horizontal half-wave dipole placed a distance h 1 above the ground level. The receiving antenna R is in the main lobe from T, at a horizontal distance D from T, and at height h2 . The signal at R is considered to be composed of the direct wave and a wave reﬂected from the ground. The reﬂection is assumed to cause a phase shift π in the wave, but no loss of power. The ground is considered ﬂat over the distance D, and h1 h2 D. 1. Calculate the electric ﬁeld E (magnitude and direction) in at R if the transmitter antenna is fed with a current j. 2. Discuss the meaning of the result.

T h1 h1

Solution

The Fourier transform of the E ﬁeld in the far zone (radiation ﬁeld) from a half-wave dipole antenna, i.e., a linearly extended current distribution with length λ 2 in the z direction,

λ 4

Since k k r and we study the radiation in the maximum direction, i.e., perpendicular to ˆ jω z J cos kz z so that k x ˆ 0, this expressions simpliﬁes to

Draft version released 15th November 2000 at 20:39

èç ê ê ë ë ë ë ê ê ë ë ê ê è äë èç ê ê ä ä ä ê ê ä ä ä ä ä ä ê ê ä ä ä ê ê é ä ä ä ê ê ä ä ä ä ê ê ä ä ä ç ääê ê ä ä

z

r1

r2

D

a

Ü e Q

ma

C) ' ì

Eω rad r

dz eik x jω

n f

Ü

i eikr r ˆ 4πε0 c r

λ 4

k

æ

ã

3. Simplify the result for the case h1 h2

) ¹a '

å

) ~¹a '

ã

W

M ICROWAVE

LINK DESIGN

E XAMPLE 6.4

λ D.

R

h2

52

L ESSON 6. R ADIATION

FROM

E XTENDED S OURCES

λ 4

A superposition of the direct and reﬂected contributions (with the distance from the transmitter T to the receiver R equal to r1 and r2 , respectively), with due regard to the phase shift π (corresponding to a change of sign in the current), gives the Fourier transform of the total E ﬁeld at the far zone point R :

2D 2D

where

and 2h1 h2 (6.48) D is the difference in path distance. Insertion of (6.47) and (6.48) into (6.46), with k ω c 2π λ , gives the Fourier transform of the ﬁeld at R

from which we obtain the physical E ﬁeld 2D

We see that the received signal at R will be extinct if

Draft version released 15th November 2000 at 20:39

7

2π h1 h2 Dλ

nπ

E ND

OF EXAMPLE

! ð3

#

ï

ì

î

C) W '

Etot r t rad

tot Re Eω rad e

D

#

1

iω t

J 2π h1 h2 2π sin cos πε0 cD Dλ λ

Ü

! Ü

e

C) ' ì

Etotrad r ω

e

∆r

#

r

D

2D

iJ eikr e 2πε0 c D

í

#

h2 1

h2 2

D

ik∆r 2

eik∆r

2

z ˆ

2π h1 h2 J eikr sin z ˆ πε0 c D Dλ h2 1 h2 2

#

#

#

#

' #

r2

h1

h2

í )

D2

2

D

D

#

)

#

h1

h2 2D

2

h2 1

h2 2

h1 h2 D

r

!

!

#

' #

!

' #

r1

h1

h2

í )

D2

2

D

D

#

)

!

' #

ã

Since h1 h2

**D, Pythagoras’ theorem gives h1 h2 2D
**

2

h2 1

h2 2

3

!

C) ' ì

# £) ' ì

d) ' ì

tot Eω rad r

Edirrad r ω

Ereﬂ r ω rad

1

iJ 2πε0 c

x Ü e ·) wu ' v Ü ¨ a

ik eikr J sin kz 4πε0 c r k 2

λ 4 λ 4

z ˆ

iJ eikr z ˆ 2πε0 c r

eikr1 r1

eikr2 r2

¨ ·)

z ˆ (6.46) h1 h2 D r ∆r 2 ∆r 2 (6.47)

'

§ w) a '

a

Ü e Q §t

~) ' ì

Eω rad r

dz J cos kz r ˆ

z ˆ

Ü

ik eikr 4πε0 c r

λ 4

r ˆ

W

ωt z ˆ

6.4

L ESSON 7

Multipole Radiation

7.1 Coverage

We look at electric dipole, magnetic dipole and electric quadrupole radiation. Multipole radiation analysis is important since it simpliﬁes the calculation of radiation ﬁelds from time-varying ﬁeld and since EM multipoles exist in many ﬁelds of physics such as astrophysics, plasma physics, atomic physics, and nuclear physics.

7.2

Formulae used

¨ ¨ ñ

ω µ0 eik x p 4π x ω 1

Fields in far regions from an electric dipole Bω rad x

Fields in the far zone from a magnetic dipole Bω rad x k

Eω rad x

**Field in the far zone from an electric quadrupole
**

Draft version released 15th November 2000 at 20:39

¨

k eik x mω 4πε0 c x

k

¨

¨

µ0 eik x mω 4π x

ó

Eω rad x

pω 1

k

k

¨

ò ® ® ¯ ¯

1 eik x 4πε0 x

ñ

® ® ¯ ¯

® ® ¯ ¯

® ® ¯ ¯

¢ ¢ ¢ ¢

k

ñ ñ ñ ñ

k

53

In what follows we use the convention that we write the dipole expression as a complex quantity but we drop the c.1) (7.4) in spherical components rather than Cartesian components since the expressions for dipole ﬁelds in spherical components are simpler. which is commonplace when discussing harmonic oscillation. stands for the complex conjugate of the term opposite the sign.c. which is due the circular rotation. It is easy to identify the Fourier component of the dipole moment in this case We notice. ˆ We would like to express (7.54 iµ0 ω eik x 8π x L ESSON 7. that the y component has the phase factor ˆ i eiπ 2 relative the x component. in this complex space variable space. This can also be rewritten in complex form or where c.. M ULTIPOLE R ADIATION 7. term. (b) Determine the radiated average power angular distribution and the total radiated power.4) .3) (7. Solution (a) We can write the time-varying dipole momentum relative the location of the dipole as which represents a constant dipole moment p0 times unit vector rotating with angular frequency ω . (a) Determine the time-dependent electromagnetic ﬁelds at large distances from the dipole.c.1 Solved examples R OTATING E LECTRIC D IPOLE An electric dipole with constant electric dipole moment magnitude is located at a point in the xy plane and rotates with constant angular frequency.3 E XAMPLE 7. so we transform Draft version released 15th November 2000 at 20:39 G # C) ' pω t p0 x ˆ ip0 y ˆ # G G # ) # ' C) ' pt p0 e x ˆ ip0 e e iω t # e î C) ' pt Re p0 e x ˆ ip0 e iω t y ˆ y ˆ ï e iω t ) # e ' C) ' pt p0 cos ω t x ˆ sin ω t y ˆ iω t õ ¢ Eω rad k Q k ¨ ¨ ¥ ô ® ® ¯ ¯ ¢ Bω rad k Q i eik x 8πε0 x ¨ ¥ ® ® ¯ ¯ k ñ ñ k F F Ü (7.2) cc (7.

19) (7.17) (7. S OLVED EXAMPLES 55 the base vectors x and y in the conventional fashion and get ˆ ˆ e Now that we have the Fourier component of the dipole moment expressed in spherical components we insert this into the dipole radiation fomulae: First we calculate from which we get p k so ﬁnally we can write the ﬁeld in space and time coordinates (remember: Bt x Re Bω x e iω t ).9) (7. So our ﬁnal expression is Draft version released 15th November 2000 at 20:39 ) ) # ! ' ! ) # ! ' ' E ˆ ϕθ sin kr ωt ˆ ϕ ϕ ) ) # ! ' # ) # ! ' ' B µ0 ω 2 p 0 ˆ sin kr ω t ϕ θ 4π cr ω 2 p0 cos θ cos kr ω t 4πε0 c2 r cos θ cos kr ù ûú) ωt Þ) ! ! ' ' h } e g Ý ! &X) W ' Bt x µ0 ω ei kr ω t Re θ p0 keiϕ iˆ cos θ ϕ 4π r µ0 ω p 0 k θ Re ei kr ω t ϕ eiϕ iˆ cos θ ϕ 4π r ) ! p 0 k 2 e iϕ ˆ cos θθ ! h e g ! 'ø ø øø ï ' e ) k p0 ke ˆ iϕ øø øø øø øø iϕ r θ ˆ ˆ 0 i 1 0 ) p0 keiϕ iˆ θ ! ø' ø øø ! $ ) p k p0 ke ˆ cos θ ϕ ˆ ϕ cos θ 0 øø øø øø øø iϕ r ˆ sin θ 1 ˆ θ cos θ 0 ) ' ! $ Eω ! $ Bω µ0 ω eikr pω k 4π r 1 eikr pω k 4πε0 r # p0 e iϕ ˆ iϕ k ˆ ϕ i 0 ˆ ϕ ϕ ÷ # # ˆ ϕ i cos ϕ ˆ r sin θ θ cos θ ˆ sin θ cos ϕ i sin ϕ Cy) ) ) iϕ ! # ) # ö ) %Cy) # 0) # 0) # # # # ' ' ' !' # # ' ' # ' y' 5) ' î pω t ˆ ˆ cos θ cos ϕθ sin ϕ ϕ ˆ ˆ i sin θ sin ϕ r cos θ sin ϕθ cos ϕ ϕ ˆ p0 r sin θ cos ϕ i sin ϕ ˆ ˆ cos θ cos ϕ i sin ϕ θ p0 sin θ cos ϕ r ˆ (7.3.13) (7.20) .5) (7.8) (7.15) (7.10) (7.14) (7.16) ' C) W ' (7.7) (7.11) (7.7.18) (7.12) (7.6) (7.

23) (7.1 . M ULTIPOLE R ADIATION monochromatic so The total power is then 0 1 1 E XAMPLE 7.26) (7.21) (7. and Q and Solution Draft version released 15th November 2000 at 20:39 ã (b) The radiation diagram when ω a c 7 ~) # ' dx 1 x2 C) # ' e Q Q µ0 p 2 ω 4 0 2π 32π 2 c µ0 p 2 ω 4 0 2π 32π 2 c µ0 p 2 ω 4 0 6π c Ω Ø Q P Sr r2 dΩ π dθ sin ω 1 ) # ' µ0 p 2 ω 4 0 1 32π 2 cr2 cos2 θ r ˆ cos2 θ ) ~y) ' ) p kk p k E ND C) ü ' ) ) ' y' ' ) ! ü ' ) ' ' ü × Ø S p k k p 1 1 µ0 ω 1 1 E B 2 µ0 2µ0 4π 4πε0 r2 ω kp k p k 32π 2 ε0 r2 ω p k 2r ˆ 32π 2 ε0 r2 ü Ø (b) We use the formula S 1 2 µ0 E × B where the E and B ﬁelds are complex and × k (7.56 L ESSON 7.25) (7. Determine (a) The Fourier components of p1 .24) (7.22) (7. The particles rotate with a constant angular speed ω in the plane of the circle. m.28) OF EXAMPLE 7.2 R OTATING MULTIPOLE Two point charges of equal charge q are located in the xy plane at either end of the diameter of a circle of radius a.27) (7.

47) (7.44) (7.40) (7.32) (7.36) (7.3.35) (7. S OLVED EXAMPLES 57 n ∞ n Fourier transform ∞ Draft version released 15th November 2000 at 20:39 ro ) ! ' ro ) ' Qω u # u qa2 δ ω 1 0 0 0 1 0 0 0 0 qa2 δ ω 2ω0 1 0 0 0 0 1 0 0 0 s ) ! ' s e Q ì Qxy ω qa2 e h e g e 1 2π ∞ i 2ω 0 ω t dt ) y) ! δ ω 2ω0 qa2 iδ ω 2ω 0 ) ! ' ! ' e Q ! ) ' ' ì Qyy ω qa2 1 2e e ∞ 1 1 2π ∞ 1 qa2 δ ω 2π 1 2ei2ω0 t i2ω0 t eiω t dt ) y) ! ' # 0) ) # ' # ' # £) ' ' e Q ì Qxx ω ∞ 1 qa2 1 cos 2ω0t eiω t dt 2π ∞ qa2 δ ω 1 2δ ω 2ω0 1 2δ ω ) ) e e ! ' # ' Qxx Qyy Qxy qa2 1 e i2ω0 t qa2 1 e i2ω0 t Qyx qa2 ie i2ω0 t 2ω 0 ) # ' ) ! ' C) e # ' Qxx Qyy Qxy Qiz q x 2 x2 2qa2 cos2 ω0t qa2 1 1 2 2qa2 sin2 ω0t qa2 1 cos 2ω0t Qyx qa2 sin 2ω0t Qzi 0 a a Qi j ∑ qn xin x jn cos 2ω0 t ) ' e Q mω z qω0 a2 ˆ 1 2π ∞ e iω t d ω qω0 a2 z δ ω ˆ ) Cy) a !' )a !' # a a ' a a m qvn qv1 r1 1 r 2∑ n n a p ∑ q n rn 0 1 r 2 1 q v1 qω0 a2 z ˆ ) ) # # !' ! a B a a ! a B a ' a r1 r2 v1 v2 a cos ω0t x sin ω0t y ˆ ˆ r1 ˆ ω0 a sin ω t x cos ω0t y ˆ v1 (7.34) (7.48) (7.31) (7.38) (7.46) (7.37) (7.42) (7.41) (7.39) (7.33) (7.29) (7.49) .43) (7.7.30) (7.45) (7.

the atom is therefore described by where cf and ce can be viewed as given constants. The power from an electric dipole is given by Solution The charge density is Only the two last terms contribute because they are non-static. the power emitted by the atom via the dipole radiation which appears due to the transition between the two states.52) E ND OF EXAMPLE 7.55) cf Ψf 2 c f ce Ψ e Ψ f c f ce Ψ e Ψ f ce Ψ e 2 (7. Determine.58 iµ0 ω eikr k Q 8π r i eikr k Q 8πε0 r L ESSON 7.59) 7 )ü ' Ø S 1 E 2 µ0 B ) ) ' y' E k ) ' B k k (7.51) × (7. chosen such that Ψ becomes normalised.56) (7.54) (7. Via inspection we ﬁnd the Fourier components for ω E 1 E0 Draft version released 15th November 2000 at 20:39 ) !' ü C) W ' ρω r θ qc f ce r cos θ exp 2r ) !' ! ' ' ü # £) ) ÿ y) ! ! ' ' !' c f ce exp i E1 E0 t c f ce exp i E1 E0 t ) ) # y) !' 4# ) !' ü þ q c f 2 r2 exp 2r cos2 θ ce 2 exp 2r r cos θ exp 4# ü ü # ü ü # þ ü &# ρ P µ0 ω 4 p 2 12π c qΨΨ # Ψ cf Ψf ce Ψ e q ) ý !' ) ý !' e e Ψf Ψe re r exp i 1 t cos θ E e r exp i 0 t E (7. according to this semiclassical interpretation.3 ATOMIC RADIATION A transition in an atom is described as quantum matrix element of a radiation operator between the un-normalised eigenstates At a certain moment.58) (7. one can interpret the magnitude squared of the wave function as a particle density function.57) 2r (7. According to semiclassical theory. M ULTIPOLE R ADIATION E XAMPLE 7.53) (7.2 .50) (7.

If we assume that we R much greater than extension x of the charge system are observing at a distance x and further that v c.71) Draft version released 15th November 2000 at 20:39 7 ! ã ã a a ºa & G ) ü ' ºa ÿ ! D! m a ! ã í a & ! P ) ! ' µ0 ω 4 p 2 12π c 16π q2 µ0 E1 3c 4 E0 4 ) Cy) ! ' ) W W ' ü # W W ' ü 12π qc f ce 0 0 1 8π qc f ce 0 0 1 8π qc f ce z ˆ 0 1 21 3 1 c f ce 2 d) ) ! ' Q # W W ' 12π qc f ce 0 0 1 π 1 2 sin 3θ sin θ dθ C) ) # ' Q 12qc f ce 0 0 2π 0 π 1 2 sin θ sin θ cos 2θ dθ C) Q QW W ' 12qc f ce 0 0 cos2 θ sin θ dθ dφ E ND OF EXAMPLE ) W W ' Q W W ' Q 1 qc f ce 4! 2 2π π cos θ sin θ sin θ cos φ sin θ sin φ cos θ dθ dφ Q Q 3 ) !' Q qc f ce r4 exp 2π 0 π 2r dr cos θ sin θ r dθ dφ ˆ ) !' ü 1 ü ü ü ü ü ü Q 2 p qc f ce r cos θ exp 2r r r r2 dr dΩ ˆ ¶ Consequently. Solution Background Consider a system of localised charges in motion.4 .60) (7.3 E XAMPLE 7.67) (7.3.7. classically by assuming that the particles travel at a constant velocity v 0 the time the annihilate.69) 7.70) x x R x R Since the timescale for the system is of the order T x v and since v c we have that ã ·a x c T (7.61) (7.68) (7.65) (7.64) (7. This is because. the corresponding Fourier component of the dipole moment p is ρω r d3x (7. due to the ﬁrst assumption x (7.62) (7.63) (7.66) (7. The power from the electric dipole radiation due to the transition between the two states is given by C LASSICAL P OSITRONIUM Calculate the radiation from a positron-electron head-on collision and subsequent annihilc up until ation. it can be argued that the source time t is approximately t R c (where R x ) instead of t x x c. S OLVED EXAMPLES 59 0 0 0 2π 0 π 0 and a similar integral for the complex conjugate term.

we rewrite the vector potential as µ0 A Σqv (7. µ0 ˙ A p (7.73) 4π R since the denominator in the integrand is now not dependent on the source coordinates.77) (7.78) (7.82) .76) 4π R Deriving the EM ﬁeld in the usual manner we get µ0 R ¨ p 4π rc2 R It can be shown that the angular and spectral distribution of energy is The Calculation The dipole electric moment of the positronium is We take the second time derivative of this ∂p 2qe v0 θ t z ˆ ∂t ∂ 2p 2qe v0 δ t z ˆ ∂ t2 For the spectral and angular distributions of the radiation we have ¨ d pω R R 2 (7.60 L ESSON 7. Thus.74) 4π R Observe that the summation can be written as d ˙ Σqx p (7. so that ˆ Draft version released 15th November 2000 at 20:39 Y Ü ) ' ) ! " ) ' ) !' ' ! " p z 2qe v0t z ˆ 0 ) ' ) y' d dω dΩ ¨ p ω R R 2π c3 ) ' B 2 ) ) ' y' E 1 ¨ p 4πε0 rc2 R R ) ! ' ) a ' Q ! t í a t R c (7.79) t t 0 0 (7. M ULTIPOLE R ADIATION and this allows us to write The vector potential is this case µ0 A j t R c d3x (7.83) dω dΩ 2π c3 where subscript ω denotes the Fourier transform of the dipole.72) R R (7.81) (7. Substituting j ρ v.80) (7.75) Σqv dt where p is the electric dipole. Now the Fourier transform ¨ of the dipole is simply pω 2qv0 z 2π 1 2 .

4 .7.3. Draft version released 15th November 2000 at 20:39 7 ) ' E ND OF EXAMPLE 7. S OLVED EXAMPLES 61 d ˆ q 2 v2 z R 2 π 2 c3 (7.84) e 0 ˆ dω dΩ In the ﬁnal result we notice that there is no dependence on ω so the spectral density is white noise.

Draft version released 15th November 2000 at 20:39 62 .

L ESSON 8 Radiation from Moving Point Charges 8.23–26).2) (8.1 Coverage ¡ ¡ In our series on deriving ﬁelds from given sources we have come to the most fundamental case: the moving point charge.1) (8.2 Formulae used According to the Formulae (F. We ﬁnd that accelerating charges radiate. We also look other mechanisms for a point charge to radiate such as Cerenkov emission. the ﬁelds from a charge in arbitrary motion are given by q Rv 1 4πε0 s3 v2 c2 ˙ Rv v 2 c Draft version released 15th November 2000 at 20:39 ¢ s x x x x v c ® ¡ ¥ ¡ ® Bt x x x Et x cx x ¤ £ ¡ ® « ¢ Et x x x ¨ ¡ ¨ ¡ ¨ ¡ ¢ « ¢ « ® (8. The ﬁelds are derived from the LiénardWiechert potentials.3) 63 . 8. In what follows we will assume that the motion x t is known “in advance”.

4) (8.11) (8.5) (8. R ADIATION FROM M OVING P OINT C HARGES x s Alternative formulae x 8.8) β cos θ (8.1 Solved examples P OYNTING VECTOR FROM A CHARGE IN UNIFORM MOTION Determine the Poynting vector for the ﬁelds from a charge q which moves with constant velocity v.7) rv r0 (8. Show that no power is emitted from the charge during the motion.10) (8. Solution In general the ﬁelds due to a single point charge may be written as # # E B Ev Bv Erad Brad ¢ ¢ £¡ ¢ ∂t ∂t s r 1 β cos θ Draft version released 15th November 2000 at 20:39 ¢ ¢ ¢ s r r r 1 β r β r cos θ ¢ ¢ ¢ ¥ ¢ Rv x x0 r rβ ¢ ¢ « β β ® ® ¢ v c ¢ ¡ x x r ¡ ¢ £ ¢ ¡ ∂t ∂t x x β ® ® ¢ Rv x x x x v c ¥¥ ¡ v c ¥¥ ¦ ¡ (8.9) (8.64 L ESSON 8.6) (8.12) .3 E XAMPLE 8.

16) (8.18) (8. or in other words β 0. ˙ It is easy to verify that for uniform motion of the charge q.19) (8. It may not be difﬁcult to get an expression for x t .14) (8. where x t is the given motion of the charged particle.13) (8.20) .3. With these last relations (8. These can be derived from the Liénard-Wiechert potentials and result in and The big outer square brackets indicate that one should evaluate their content at the retarded time t t x x t c.8. that 0 and that 1 B β E c so the Poynting vector in this case is Furthermore.15) (8. S OLVED EXAMPLES 65 Ev and Bv are known as the velocity ﬁelds and Erad and Brad are known as acceleration ﬁelds. but then to solve the equation for the retarded time t t x x t c for t t t x . it can be shown that where θ is the angle between r0 and β . Inserting this relation for E into the relation for S we obtain Draft version released 15th November 2000 at 20:39 Ü ) ! ' ) ! ' E q 1 4πε0 β2 3 r0 1 1 β 2 sin2 θ r 3 2 0 ! E where we have introduced the virtual position vector r0 we may write ) ' ! C) ' ! s r β r 2 r0 r0 β 2 ) ' C) | ' S β E β 1 E µ0 c E Erad Brad ε0 E µ0 E r rβ .21) E ) W 'a a ) ra ' a ) ra ' a )ra ' a ) a ' a ! ! ! "! ) a a ' ! s r β r E where β v c and (8. In this case it is not necessary to perform this transformation of variables since we are not interested in the time evolution. It is this that makes the equations for the ﬁelds difﬁcult to evaluate in general.17) ) ! ' Ý Brad r Þ rβ ! Bv q β r 1 4πε0 cs3 q r r ˆ 4πε0 c2 s3 β2 ˙ β ) ! ' Erad ·) ˙ β ! ' ) ! ' Ev q r rβ 1 β 2 4πε0 s3 q r r rβ 4πε0 cs3 (8. so we drop the brackets.

so a charge in uniform motion in vacuo. So that Draft version released 15th November 2000 at 20:39 ) ˙ v r r rβ ˙ v r2 r2 β cos θ ˙ vr2 1 β cos θ ˙ vrs ) ) ¨ ·) ' ' r rv ˙ v ˙ rv r v ˙ v r rv ! & ! ' ! & ! ' & ! ! ' § & ! 8 v ! ) x wu ' t X) ˙ Now.28d) 7 ) ' Ò Ò P S dA ˆ ˆ S θ r dΩ ) ! " S ) ! ' ! ' q2 v1 4 16π 2 ε0 r0 1 ! "~) ' r0 β β r0 ˆ β r0 sin θ ϕ 2 ˆ β r0 sin θθ r0 β 2 2 sin θ ˆ θ β 2 sin2 θ 3 0 ) ! ' 6 r0 1 ) E ND ' ) ! ' S β2 2 r 3 0 β ε0 q 2 1 2 µ0 16π 2 ε0 1 β 2 sin2 θ r0 (8. Solution We consider only the formulas for the radiation ﬁelds.27) (8. we have that v r 0. so ) ! ' ! ! s r r β r rβ cos θ r 1 β cos θ (8.1 . E XAMPLE 8.28b) (8.26) OF EXAMPLE 8.24) (8. for which the denominator is the cube of the retarded relative distance 0 where θ is the angle between the velocity and r. does not radiate energy.66 L ESSON 8.28c) (8.23) (8.28a) (8.25) (8.2 S YNCHROTRON RADIATION PERPENDICULAR TO THE ACCELERATION Determine the angular distribution of synchrotron radiation in the plane perpendicular to ˙ the acceleration v for a charged particle moving with velocity v.22) (8. R ADIATION FROM M OVING P OINT C HARGES So and So ﬁnally And now when we integrate the Poynting vector over a spherical surface A with radius R which encloses the moving charge which results in since the Poynting vector is not radial.

i.3 .29) 4 (8.8. but it can also be seen as an electric dipole relative a co-moving coordinate system. S OLVED EXAMPLES 67 ˙ µ0 q µ0 q vr ˙ vrs 4π s3 4π s2 The Poynting vector is given by Remember that the Poynting vector represents energy ﬂux per unit time at the ﬁeld point at the time t. It can be derived from the radiation ﬁelds of a non-relativistic (v c) charged particle.32) a ) ! ' a S 1 ˆ E2r µ0 c 2 v2 r ˙ ˆ µ0 q 2 2 µ c r2 1 16π 0 β cos θ ! " ! " Erad (8. The charge’s energy loss must be related to the time t .3.e.34) 7 ) ! ' ) ) ! ' ! ' ã a ∂U ∂t ∂U ∂t Sr r 2 µ0 q 2 ∂ t ∂U 1 β cos θ ∂t ∂t 16π 2 c r2 1 µ0 q 2 v 2 ˙ 1 16π 2 c 1 β cos θ 3 v2 ˙ β cos θ 4 r2 (8.31) E ND OF EXAMPLE 8. We shall investigate the latter.30) (8. the time when the energy was emitted! but T HE L ARMOR FORMULA Derive the Larmor formula by calculating the radiated power of an accelerating point charge due to electric dipole emission.33) (8.2 E XAMPLE 8. Solution The Larmor formula is a very useful equation for deriving the power of emission from non-relativistic accelerating charged particles.. Apply the Larmor formula to linear harmonic motion and circular motion. First derive the power emitted by an electric dipole. Using a time domain (non-Fourier) version of the dipole ﬁelds: µ0 ¨ ˆ p r 4π rc we ﬁnd the Poynting vector to be ) ' B Draft version released 15th November 2000 at 20:39 ) ' E r ˆ 1 ¨ p 4πε0 rc2 r ˆ (8.

R ADIATION FROM M OVING P OINT C HARGES 0 0 0 2π π 0 ¨ p 2 sin3 θ dφ dθ 0 ¨ p2 4 8πε0 c3 3 ¨ p2 6πε0 c3 Linear harmonic motion q 2 4 ω0 x2 0 Circular motion Draft version released 15th November 2000 at 20:39 W) ' C) ' W ) ' C) ' xt Ø P C) ' Pt cos ω0t 6πε0 c3 2 4 q 2 ω0 x2 0 12πε0 c3 R0 cos ω0t yt R0 sin ω0t G) ' ! "C) ' ) ' C) ' W ) ' C) ' xt x0 cos ω0t at xt ¨ 2 ω0 x0 cos ω0t ) ' E C) ' where we have identiﬁed the acceleration a t C) ' Pt ¨ q2 x 2 6πε0 c3 ) ' C) ' pt qx t Q ¨ p2 8πε0 c3 π 0 sin3 θ dθ q2 a 2 6πε0 c3 ¨ xt .68 L ESSON 8.35) (8.37) (8.38) (8.36) (8.40) (8. ¨ p Q Q Q Q 1 16π 2 ε0 c3 1 16π 2 ε0 c3 Q Q 25) ' Pt S r r2 sin θ dφ dθ ˆ π 2π 0 π 2π ) ' So integrating for the power P t r 2 sin θ dφ dθ ˆ ) ¨ p ' ) C) r ˆ (8.42) ' ' ¨ ! ) '§ S 1 ¨ ˆ ¨ ˆ p r p r r ˆ 16π 2 ε0 r2 c3 1 ¨ ˆ ˆ ˆ ¨ ˆ p r 2r r p r 2 ε r 2 c3 16π 0 1 ¨ ˆ ˆ p r 2r 2 ε r 2 c3 16π 0 1 E µ0 B × .39) (8.41) (8.

44) 2 2vx t t y2 0 2vx t t y2 c2 t n2 t ) ! þa ' ) Cra ' a The motion of the charge is given by x t vt t so that (8.8. It remains to show that the result is physically reasonable. Solution x2 v2 t t or. And if the charges are distributed evenly but ﬂuctuate thermally then power is proportional to N. Draft version released 15th November 2000 at 20:39 # # £) a ! ' ) ³a ! ' # )a ! ' 3 ! v2 c2 n2 t t 2 2vx t t y2 x2 )a ! ' ! # £) a ! ' # )a ! ' 1 a # x2 v2 t t )a ! ' ! r2 c2 t n2 )a ! ' F but r c n t t is the retarded distance so that t 2 2 # 0) a ! ' # )a # ·) ! a ' ! $ ¨ § ! ' # )a # ! ' r2 x x 2 y2 x vt t 2 y2 (8. this is incoherent radiation.45) 2 0 0 (8.4 . from the expression for r 2 above: This is a quadratic equation in t t . this is coherent radiation.3 E XAMPLE 8.C ERENKOV EMISSION ˇ Show that the potentials at time t at a point inside the Vavilov-Cerenkov cone receive contributions from exactly two positions of the charged particle. What must be understood is whether or not the sources are radiating coherently or incoherently. One point that should deﬁnitely be raised is power radiated for more than one accelerating charge. For example.3. S OLVED EXAMPLES 69 Validity of the Larmor formula The Larmor formula although not covariant in form can indeed be extended such as to be valid for all inertial frames. For a ﬁxed t we would in other words have two values of t .46) (8.47) 7 ~) ' Pt 4 q 2 ω 0 R2 0 6πε0 c3 (8. the radiation is not automatically proportional to N. consider the above case of circular motion.43) E ND OF EXAMPLE 8. It is not so simple that one may assume that power is proportional to the number of sources N. If we had a large collection of particles such as the case with in a particle storage ring or in circular wire. ˇ VAVILOV. If they are bunched the power is proportional to N 2 . If the charges are distributed homogeneously the radiated power is 0.

x y is inside the cone! It remains to be shown that x 0 0 0 Draft version released 15th November 2000 at 20:39 7 xI ! v wu ! Ht x 3 vwu ! © t )a ! ' t t v2 c2 n2 vx 1 cos2 αc cos2 α 1 I ! ! vx 1 cos2 αc cos2 α 1 E ND OF EXAMPLE I ! vx ) ! ' ) ! ! ' 1 cos2 αc 1 cos2 α cos2 α I ! ! vx sin2 αc 1 cos2 α tan2 α ! ! vx ) c2 x2 y2 n 2 v2 x2 ! ! ) # ' # ' H H H 3 ! 1 )a ! ' t t v2 c2 n2 c2 2 x n2 y2 v2 y2 y2 x2 ) W ' Y 5) # Y # Y ' V F F F F V § ¨Ô ) a 1 ! ' t t c2 2 x y2 v2 y2 n2 c2 y2 n 2 v2 x2 y2 2 sin αc sin2 α α αc ! v2 c2 n2 ! ) # ' vx c2 n2 x2 y2 v2 y2 (8.48) ) ! ' | ! 5) a ! ' t t vx vx 1 ) (8.49) (8.54) 8.50) (8.52) 1 (8.53) (8. R ADIATION FROM M OVING P OINT C HARGES ˇ where αc is the critical angle of the Cerenkov radiation. it is half of the opening angle of the shock wave of the radiation.51) (8. So that.70 L ESSON 8. in other words.4 ! ') # ' ! 2vx 4v2 x2 4 x2 y2 v2 2 v 2 c2 n 2 c2 n 2 ! .

1 Coverage The generation of EM ﬁelds via Liénard-Wiechert potentials are considered simultaneously with the Lorentz force to give a self-consistent treatment of radiation problems.1) 71 . We also discuss the effect of radiation on the motion of the radiating body itself known as radiative reaction 9.L ESSON 9 Radiation from Accelerated Particles 9. In the previous lesson we solve radiation problems from given expressions for the motion.2 Formulae used qE v B The covariant Lorentz force Draft version released 15th November 2000 at 20:39 ¨ ¢ d mγ v dt (9. Now we consider how charges actually move in the EM ﬁelds and thus present the Lorentz force.

so the Lorentz force becomes q dv E v B (9. (i. v is the three-velocity. the Newtonian force deﬁnition). So what happens in a combined electric and magnetic ﬁelds? The equations of motion are given in (9.3 E XAMPLE 9. Hint: Separate the motion v into motion parallel and perpendicular to the magnetic ﬁeld respectively.72 L ESSON 9. The fact that the Lorentz force in this form d mγ v qE v B (9. and in a static and homogenous magnetic ﬁeld the charged particles perform circular motion. In this certain sense the Lorentz force is trivial: it is simply a deﬁnition of the EM ﬁelds.5) dt m This equation is the equation of motion for non-relativistic charge particles in EM ﬁelds.3) F dt where m is the mass of the particle.e. The equation for the Lorentz force is relativistically correct as it stands. In many cases one has conditions which are non-relativistic and under such conditions it is possible to simplify (9. as long as one interpretes d mγ v (9.5) The ﬁrst step in solving this equation is to separate the motion into motion parallel with the B ﬁeld which we will denote with v and Draft version released 15th November 2000 at 20:39 9 ! ) ' E ) ) ) # ' # ) # ' ) ' ' F qE v ' 9 B (9. The E ﬁeld gives the force parallel to direction of motion and B gives the force perpendicular to the direction of motion. γ 1 1 v2 c2 .4) dt is Lorentz invariant is not immediately clear but can easily be shown.4). The motion As is well known. R ADIATION FROM ACCELERATED PARTICLES 9.2) . On the other hand this equation is difﬁcult to solve for v t because γ contains v.3) and thus also (9. Solution Background on equations of motion for charged particles As we know. charged particles are uniformly accelerated in a static and homogenous electric ﬁeld. and t is the time. the motion of a charged particle in electric and magnetic ﬁelds is given by the Lorentz force This can be seen as a deﬁnition of the E and B ﬁelds and also the fundamental equation for measuring the ﬁelds. One simply uses the fact that as v c 0 then γ 1.1 Solved examples M OTION OF CHARGED PARTICLES IN HOMOGENEOUS STATIC EM FIELDS Solve the equations of motion for a charged particle in a static homogenous electric ﬁeld E and magnetic ﬁeld B. In this case F mdv dt.

Interpretation of the Motion Having derived the solutions for the motion of the charged particle we are now in a position to describe it in words. so that part of the E ﬁeld which is along the B ﬁeld accelerates the charge as if there were no B ﬁeld. First we have the motion along the B ﬁeld which simply is not effected by the B ﬁeld.3. With this assumption.11) (9. First of all we see that the motion consists of three separate parts.12) (9.6) # (9. Let us call the solution to the inhomogeneous equation ω D and the solution to the homogenous equation ωR .7) F F ! (9. It is also inhomogeneous which means that we have both a solution to the homogenous equation and solution to the inhomogeneous equation. which here we have denoted ω r . The motion perpendicular to the B ﬁeld can further be separated into two parts. so for example c 1 i 0 if B is in the 3 direction.9) dt m m This is a ﬁrst order linear ordinary differential equation in v . One part. and that is left is ωD B E . so v v v and thus (9.10) (9. S OLVED EXAMPLES 73 q E dt m and for the perpendicular direction q E v B (9. we notice that neither the inhomogeneity nor the coefﬁcient of the zeroth order term depends on t so the solution ωD itself cannot be time dependent. represDraft version released 15th November 2000 at 20:39 ) W W ' e R d ωr dt ωD q ωr m E B2 0 B B ωD B B E ωD B B B 2 ωD B E B ) # # ' ) )x 8 vwu ' t ¦F ! ' ! ! dv dt dv ) # ' # y { dv dt dt and since the and dv q q E v B E m m are mutually orthogonal the equation separates into # y motion perpendicular to B. can be shown to have the form ωr c e iω t where c is a constant vector perpendicular to B and satisﬁes c ic 0 (where the denotes scalar product deﬁned as the inner product of the vectors).9. This is easily solved. For the inhomogeneous solution. by taking the cross product of this equation with B we ﬁnd The solution to the homogenous equation on the other hand. which we will denote v .8) m dv q q v E B (9.13) .

74

L ESSON 9. R ADIATION

FROM

ACCELERATED PARTICLES

y x

z

Figure 9.1. Motion of a charge in an electric and a magnetic ﬁeld. Here the electric ﬁeld is along the y axis and the magnetic ﬁeld is along the z axis. There is a background velocity in the z axis for reasons of clarity. ents so called gyro-harmonic rotation, i.e. the particle moves in a circular orbit, with its axis of rotation parallel with the B ﬁeld with period of rotation 2π ωc where ωc qB m is known as the cyclotron frequency. This gyro-harmonic motion is the complete solution to our problem if the E ﬁeld was not considered. The second part of the motion in the perpendicular plane however, which we here denoted ωD involves both the B ﬁeld and the E ﬁeld. It is known as the drift velocity since this is the velocity of the centre of the gyro-harmonic motion. The motion is illustrated in Figure 9.1.

E XAMPLE 9.2

R ADIATIVE REACTION FORCE FROM CONSERVATION OF ENERGY Correct the equations of motion for a charged particle in EM ﬁelds to include the effects of the energy loss due to radiation emitted when the particle accelerates. Assume nonrelativistic conditions, so radiated energy is given by Larmor formula.

Solution

Background Among other things, Maxwell’s introduction of the electric and magnetic ﬁelds was a mathematical technique to divide the work of solving the motion of charge particles in EM ﬁelds. Instead of action-at-a-distance, the ﬁelds naturally divide the probDraft version released 15th November 2000 at 20:39

7

E ND

OF EXAMPLE

9.1

9.3. S OLVED

EXAMPLES

75

lem into two parts: the generation of ﬁelds from moving charged particles and the motion of charged particles in EM ﬁelds. In a certain sense this can be seen as a division into, on the one side a cause and on the other side, an effect of EM interaction, but the formulas governing these processes are not very symmetric with respect to each other in classical electrodynamics: the generation of ﬁelds from moving point charges is determined via the Liénard-Wiechert potentials and the motion of point charged particles is determined by the Lorentz force. Despite the lack of symmetry, this division seems successful at describing EM interaction except for one thing: it fails to describe the self-interaction of charged point particles. We know that the motion of charged particles is governed by the Lorentz force, but at the same time we know that acceleration of charged particles causes energy emission. These two facts have been, until now, treated separately, but taken together we realize something is missing. This can be seen for example in the case of a single charge under the inﬂuence of a mechanical force in a region of space with no EM ﬁelds except for the ﬁeld from the charge itself. From the Liénard-Wiechert potentials we know that the mechanical force will cause the charge to radiate and if energy is to be conserved the emission must take its energy from the kinetic energy. Since the Lorentz force is zero and there are no other electromagnetic interactions that we know of, as of yet, we have no way of accounting for this radiative “friction”. This, as of yet not mentioned force, is known as radiative reaction or the radiative damping force. One question that comes to mind, after the above discussion, is how can so many problems be described by classical electrodynamics without considering radiative reaction? Obviously, it should have a negligible effect in most cases but what are the limiting conditions? Certainly, these should be determined by considering the conditions under which the energy emitted is of the same order as the kinetic energy of the charge. If we consider non-relativistic motion, the energy emitted by a charge accelerating at the order of a, under a period of duration of order T , is given by the Larmor formula and is of the order of

On the other hand, the acceleration bestows the charge with kinetic energy on the order of

or T q2 3πε0 c3

)

'

ã

q2 a2 T 6πε0 c3

m aT 2

2

ã

So if we demand that Erad is equivalent to

Ekin

)

'

Erad

q2 a2 T 6πε0 c3

(9.14)

m aT 2

2

(9.15) Ekin and we wish to neglect the radiation reaction, then this

(9.16)

(9.17)

Draft version released 15th November 2000 at 20:39

76

L ESSON 9. R ADIATION

FROM

ACCELERATED PARTICLES

and if we deﬁne the characteristic time as τ q2 3πε0 c3 we can say that the effects of the τ. radiative reaction are negligible in measurements made over timescales on order of T

**Accounting for radiative reaction Having demonstrated the need for a force which
**

accounts for radiative effects of accelerating charges we set out to determine its form. From the conservation of energy it is clear that the force we are looking for, which we will denote Frad , must satisfy

where have integrated the Larmor formula over time on the right hand side. Partial integration yields

if assume periodic motion then we ﬁnd that

and so

We now can correct the equation of motion to include the radiative reaction

This equation is know as Abraham-Lorentz equation of motion. Unfortunately, the Abraham-Lorentz equation of motion is not without its own inherent problems.

Draft version released 15th November 2000 at 20:39

7

~)

! '

˙ mv

¨ τv

Fext

Frad

¨ q2 v 6πε0 c3

¨ mτ v

3

!

Frad

¨ q2 v 6πε0 c3

v dt

0

!

Q

Frad v dt

¨ q2 v v dt 6πε0 c3

˙ q2 v v 6πε0 c3

! Q "

Frad v dt

˙ ˙ q2 v v dt 6πε0 c3

(9.18)

(9.19)

(9.20)

(9.21)

(9.22)

E ND

OF EXAMPLE

E

1

Q Q Q

9.2

24) (9. To ﬁnd the expression for the power loss due to radiation we need an expression for the accelera- Draft version released 15th November 2000 at 20:39 3 b») ' # v ˙ m0 γ v v 1 dp dt γ2 2 ˙ v v v c2 0 a a ˙ But v v 0. disregarding radiation losses.9.28) . and p is the space part of the (9. the circular motion does emit EM radiation. However. so C) 1 3d γ v v 2c2 dt γ3 ˙ v v c2 a I ! 1 v 2 c2 3 ! 1 ! " ' H a dγ dt d dt 1 1 3 γ 2 d v2 dt c2 a a W # G ) a particle’s total energy. is which comes as no surprise.23) 0. S OLVED EXAMPLES 77 An electron moves in a circular orbit in a synchrotron under the action of the homogeneous magnetic ﬁeld B0 . since we know that the electron moves in a circular orbit. (a) Calculate the energy which is lost in electromagnetic radiation per revolution if the speed v of the electron is assumed to be constant. if B0 C) ' 'a a a a R ADIATION AND PARTICLE ENERGY IN A SYNCHROTRON E XAMPLE 9. (b) At which particle energy does the radiated energy per revolution become equal to the Hint: d m0 γ v dt Solution (a) Since we wish to consider synchrotron motion we use the relativistically correct equation of motion dp qv B dt where the right hand side is the Lorentz force with E momentum 4-vector so dp dγ d ˙ m0 γ v m 0 γ v m0 v dt dt dt where dp ˙ m0 γ v (9.26) dγ dt 0 and thus (9.25) (9.27) dt This means that the power necessary to keep the particle in a circular orbit.3.3 1 5 T? dU dt ˙ µ0 q 2 v 2 γ 4 6π c F (9.

31) (9. that is This equation is easily solved to give The factor m γ0 ωc is known as the synchrotron angular frequency and is the relativistic 0 value of the angular frequency for gyro-harmonic motion or cyclotron angular frequency.78 L ESSON 9.32) 2π ω c .37) µ0 q 3 B 0 3cm2 0 1 µ 0 q3 B 0 3cm2 0 2cm2 0 µ0 q 3 B 0 (9. R ADIATION FROM ACCELERATED PARTICLES ˙ tion v . To this end we use what is left of (9. we need the period of revolution which is T so 2 dU 2π µ0 q2 ωc γ 4 2 v dt ωc 6π c µ0 q 3 γ 3 B 0 v 2 3cm0 2 µ0 q 2 ω c γ 4 2 v 6π c (b) The task here is to equate the total energy and the radiated energy and solve for velocity and then see what total energy that velocity is associated with. So.29) B all we (9.36) (9. . Thus the particle energy for which the radiated energy is equal to the total particle energy is # ª γ 1 # v2 c2 1 ! c2 v2 c2 1 v2 γ 2 µ0 q 3 B 0 3cm2 0 1 (9. after a little algebra. Now we can insert this expression into the relativistic generalisation of the Larmor formula 2 ˙ µ0 q 2 v 2 γ 4 µ0 q 2 ω c γ 4 2 dU v dt 6π c 6π c To ﬁnd the energy per revolution.30) (9.34) µ0 q 3 γ 3 v2 B0 3cm0 . E qB v ˙ & ˙ m0 γ v q vB0 q B0 v m0 γ The acceleration is found by taking the norm of this last equation and since v have is the scalar equation ˙ m0 γ v qv a F F F B (9.33) (9.23).38) Draft version released 15th November 2000 at 20:39 ) ' The total energy is E m0 γ c2 and the radiated energy is Urev which when equated gives a Urev T (9.35) (9.

43) v c (9. Show that the radiation loss is negligible. Assume that under all circumstances E0 108 V/m. compared to the particle’s own energy gain.42) (9. R ADIATION LOSS OF AN ACCELERATED CHARGED PARTICLE A charged particle. even at relativistic speeds.40) (9. Solution The relativistic equation of motion is ˙ Having found v we try to derive the radiation ﬁeld generated by this motion. initially at rest.3 F F E XAMPLE 9. is accelerated by a homogeneous electric ﬁeld E 0 .39) E ND OF EXAMPLE 9.44) 7 V G # ª a E m 0 c2 1 2cm2 0 µ0 q 3 B 0 0 16 TeV (9.9.4 .3.41) (9. S OLVED EXAMPLES 79 This is obviously the upper limit for which radiation effect can be neglected in the treatment of synchrotron motion of the particle. we have that 3 3 Draft version released 15th November 2000 at 20:39 ) µ0 q 1 ˙ r v r 4π r3 1 β cos θ 3 µ0 q 1 ˆ r2 v sin θθ ˙ 3 1 4π r β cos θ 3 µ0 qv ˙ sin θ ˆ θ 4π r 1 β cos θ 3 ! ro r r ˙ v 8 u x wv u t ' ) ) ) ) ! ' ! ' µ0 q 4π r3 1 1 β cos θ r v c ˙ v 0 s ! ß r r r ˙ v ) ) ' ! ' ! ' ! ' Erad v ˙ q E0 m0 γ 3 µ0 q ˙ r rv v 4π s3 µ0 q 1 4π r3 1 β cos θ & ˙ m0 γ 3 v q E0 ! 1 β2 & D3 m0 γ v ˙ 1 m0 γ vγ 2 ˙ q E0 3 ! # C) 1 # ' ˙ m0 γ v 1 β 2γ 2 m0 γ v 1 ˙ 1 Since v ˙ v E0 then β2 1 β2 3 D¸) ' # ˙ m0 γ v 1 C) 'a dp dt d m0 γ v dt γ2 ˙ vv v c2 qE0 (9.

51) .48) m 0 γ c2 . R ADIATION FROM ACCELERATED PARTICLES where have used the particular geometry of the vectors involved. which is The radiated energy per unit area per unit time corresponding to this is Sr r 2 but the energy radiated per unit area per unit time at the source point is The total radiated energy per unit time is ˜ ∂U ∂t From the equation of motion we ﬁnd that Hence m0 γ v ˙ q E0 γ2 Draft version released 15th November 2000 at 20:39 & d dγ m0 γ v m0 γ v m0 v ˙ dt dt and.45) r2 (9. Then we may determine the Poynting vector.80 L ESSON 9. so (9.47) (9.46) rs Erad µ0 c 2 6 (9. from the expression for v.50) (9.49) (9. ˙ a # a ~) dE dt m0 c2 dγ dt q E0 We compare this expression for the radiated energy with the total energy E 3 µ0 1 x2 8π 2 c 1 βx 5 1 βx 5 1 2 2 ˙ ˙ µqv 4 µ q2 v2 4 6 1 2π 0 2 2π 0 2 γ 2 3 8π c 3 1 β 8π c 3 ) # ' ! ) ) # ' ! ' e Q 2π dx 2 µ0 q 4 E 0 6π m2 c 0 ) # ' 1 dx sin θ q2 v2 1 ˙ ! e Q x dθ cos θ ) ! ' Q ∂U dΩ ∂t ˙ µ0 q 2 v 2 2π 16π 2 c ) µ0 q 2 v 2 ˙ sin2 θ 2c 1 16π β cos θ 5 sin3 θ dθ 1 β cos θ 5 0 µ q2 v2 1 1 x 2 ˙ 2π 0 2 dx 8π c βx 5 1 1 π ) ! ' ) ! ' !" a ! ' r2 1 2 ˙ β cos θ µ0 q2 v2 sin2 θ µ0 c 16π 2 r2 1 β cos θ øø øø a ( Q ∂U ∂t ∂U ∂ t ∂t ∂t ∂U s ∂t r ø µ0 c øø øø ø ∂U ∂t Erad 2 øø øø 'a a a a S 1 2 ˆ Erad r µ0 c Sr r ˆ (9.

and an electron for which q & m0 c2 dE dt dγ dt q E0 v m0 c2 q E0 v m0 c2 q E0 v e.55) 10 8 m/s) 9.3. for E0 108 V/m. S OLVED EXAMPLES 81 v2 c2 and 31 2 11 10 Draft version released 15th November 2000 at 20:39 7 E ND OF EXAMPLE G Which of course is a very small ratio for a relativistic electron (v even for E 108 V/m.54) v (9.4 3 v wu ! t & a m0 v q E0 1 1 dγ dt 1 γ2 (9.53) (9. c 2 998 e G ) ) E0 4π 10 7 1 6 10 19 v 6π 2 998 108 9 11 10 e e G ' G ' ) e G ' a a ˜ dU dt dE dt 2 µ0 q 4 E 0 1 6π cm2 q E0 v 0 µ0 e 3 E 0 6π cm2 v 0 3 12 E0 Finally. we ﬁnd that x (9.9.52) a a F .

82 L ESSON 9. R ADIATION FROM ACCELERATED PARTICLES Draft version released 15th November 2000 at 20:39 .

- Tool Monitoring Using Image Processing
- X2_Text recognition.pdf
- Mathcad - Pond Hex_final
- X2 Thesis Ibrahim Alenezi
- Mutation 2
- Mutation 1
- Bell Delaware Math Cad Example
- ME410 Labsheet_Exp4_ Fall 2011
- Me400 Ozet Notlar a Cinar
- Workplace and Labor Health
- Practical Training on Arduino
- C# Interview Questions
- 4th Year Courses
- Introduction to Basic Electronics
- c++ - Generate All Subsets of Size k From a Set - Stack Overflow
- Animation of Scotch-yoke
- ME308_Tutorial8_2012
- ME308_Tutorial7_2012
- ME308_Tutorial5_2012
- ME308_Tutorial9_2012
- ME307-11 Tutorial 9
- ME307_11_Tut11
- Kayit Kilavuzu 2011 Eng
- production of ethanol
- Transformation of Ottmn

Sign up to vote on this title

UsefulNot useful- Concepts Electro mby Prashanth Shyamala
- paperby bhatiamanavgmail
- 1873-7005_45_1_015505by Kristen Charles
- On some equations concerning Fivebranes and Knots, Wilson Loops in Chern-Simons Theory, cusp anomaly and integrability from String theory . Mathematical connections with some sectors of Number Theoryby Michele Nardelli

- zEMFT Thidé Book Exercises Eletromagnetic Field Magnetic
- EMFT Exercises
- Electromagnetic Field Theory Exercises- Bo Thide, Upsilon Books 2001
- ma4k5
- Classical Mechanics.pdf
- Vector Wave Equation
- 1311
- 2d_wells
- Concepts Electro m
- paper
- 1873-7005_45_1_015505
- On some equations concerning Fivebranes and Knots, Wilson Loops in Chern-Simons Theory, cusp anomaly and integrability from String theory . Mathematical connections with some sectors of Number Theory
- 0901.0853
- Enigma of Non-holonomic Constraints
- Lecture 8
- Paper 4
- Conductor2.pdf
- Transport Phenomena
- 0202109
- Heinbockel - Tensor Calculus - Part
- Statics note
- Noninertial Frames
- symetries
- QFT notes 3
- 1Q09 Influ Coeff
- Linear_Equations
- 0009103
- fdfd_waveEqn
- basiceqs.pdf
- UM notes
- Electromagnetic Field Theory Exercises- Bo Thide, Upsilon Books 2001