‘MSci 4261 ELECTROMAGNETISM LECTURE NOTES Il Jn this section of the notes we will discuss the simplest light: waves (plane waves) in simple non-conducting media and review the concepts of polarisation and dispersion. 3.1 Plane Waves Recall Maxwell’s equations in the absence of sources - V-D=G, _ oD vxHa®, V-B=0, _ _JB VxB=-F Assure we have uniform isotropic near media, for which D = 48, = Hf, and write the dectric and magnetic fields as EG) Ect, Bex t) = Bye, where Ey ane Bo are constant vectors and mis a constant unit vector. Then Maxwell's equations are satisfied provided that (Exercise: prove this) B= pa, n-Ey=0, n-By=0; Bp = Vien xB) (we scale k so that: m & of unit length). ‘These conditions mean that firstly, the electric and magnetic fields are perpendiclar to the direction of propagation of the wave n, ie this is a transverse wave. Furthermore, the dlectric and magnetic fidds are perpendicular to each other (from the last equation). Also, B and B have the same magnitude for these waves in the case of free space (for which flees = 1/c). This follows from taking the vector nom of the ks. equation, and wing the above facts. Its straightforward to show that the energy flow per unit area par unit time given by LE x B = /e/pi[Bo + Eo|n and that the time-averaged exergy density is $(€E° + 1B?) = efEp - Eo), with the speed of energy flow thus being v = 1/y/f; the phase vdocity. (The phase Velocity is defined to be v= #f = 1/ VR@=c/n, and the index of refraction n is n= (5) 3.2 Polarisation Using the results of the section above, we introduce a set of real mutually orthogonal unit vectors (e1,e2.), where e; is paralld to Ep and e, is paralld to Bo. Then we may write the solution discussed above as E=ek, B=evfel, where Fy = [Eo)- We note that we may have altematively written E=eh, B=-evieiy as another sohution of Maxwell’s equations (this is just a 90° rotation of the first solution, with a different constant £4). The first solution above is described as a Hinaarly polarised wave, with polatisation vector 1, singe the electtic field is always in the e1 direction. Note that, by definition, the plane or direction of polarisation of a wave equals the plane or direction of the electric fied E. Similarly, the second solution is linearly polarised with polarisation vector e2. We can combine the two plane waves to form the most general homogeneoxs plane wave propagating in the direction k = km, B(x, f) = (1B + eka)", 1with the amplitudes F,,F2 constants. Allowing these constants to be complex permits phase differences between waves of different polarisations. For example, consider the solutions B50) = Bi(er + ies", ‘The physical dectric field (which is defined here to be the real part of the solution) is then constant in magnitude, but swears out a citck with frequency w as time evolves (ata given point in space). This is because REE = Eye; cos(t) + Exe. sin(ut). This is a circularly polarised wave (with positive helicity, since the rotation is dockwise for an observer facing the oncaming wave). In general there is a four parameter family of polarisations, described by the Stokes parameters. These describe the helcities (positive and negative) and the linear polarisations in the two directions papendicular to the waves’ motion, 33 Dispersion Dispersion means that the effect of a medium on light traversing it depends on the frequency of that light. In the above we implicitly assumed that the parameters p, € were real and constants. This is thus the dispersion free case. As we saw, in this case, waves travel without distortion. In reality, all matter exhibits dispersion, with different frequencies of light being affected differently by the media, As a simple model for the physics of dispersion, consider the modd of matter as molecules localised at sites, An applied electric field produces an induced electric dipole moment. in each mokeaule. Suppose that the dectrons in cach molecule are bound under the action of a restoring force F = —mujfx, with m the mass of the charge and wp the oscillation frequency. The equation of motion for the electron is @x dx e BB tue E(x, 1) im (7is the damping force, ~e the charge on the electron, the relative permeability has been taken to be one, and this is for small oscillations and magnetic forces). Assuming time dependence of e~#", ie x= xye™, ‘we have the dipale moment far an electron as new = Egat in Hence the didectric constant € = (1 + Xe)éo is given by = 04S (Recall that: p= ¢oxeE defines Xe.) This fornmila is the basis for a good description of the atomic contribution to the didlectric constant. Note that this €is now dependent upon the frequency w ico % called the resonant frequency. ‘Typically; is small in compertson to the frequency wp, $0 that the imaginary part of ¢ is small. In the normal case there is more than one resonant frequency (eg due to more than one electron or molecule). ‘Then, for frequencies below the smallest resonance, ¢/«) greater than one— if wy is small, then fw 0 we must close the contour in the lower halplane to get. convergence on the are at infinity, and the contour then encloses the poles of the integrand (in the negative sense, which introdueas an extra minus sign), giving with use of the residue theorem 2 ur 2 {So ents ony G@)= for 7 >, Putting these results together a ( GQ =wperP RID. 6) ‘The details are not s0 important as the presence of the factor 6(7) which is ckerly still present even in a more sophisticated modd in which there would be a sum over similar terms (many oscillators, each contributing aterm as above with an oscillator strengfi), or ieleed in the full quantum mechanical treatment (which in fect gives an explanation of the oscillator strengths). ‘There is a ‘non-locality’ in time in the relation between D and E by of order 77! = 10-8 sec. for a typical spectral line; -y is the natural line width. The factor 6(r) is essentially independent of the model, and it ensures that the respons of the dielectric is causally related to the E-field, 1 Bd0«.9 = Ban + [* GB E&t— dr ssince the lower limit of the 7-integration may be taken as 0, not -o0. We have G(r) ~ e~7", with + acting Tike a damping force. For 7 > 1/7, G is damped, and for times up to T ~ 1/7, E(t—7) affects D(t). This then means that _ ele) <4 [ Ge dr lo (this is just the inverse Fourier transform). Because G(r) i rh i is easy to see that el) = flu I We also we that ¢ is an analytic function of w in the upper aéplane (fam equation (*) above), The analyticity of e,() may be extended to Su > 0. As ¢- — 1 is analytic, the Cauchy theorem then allows us to wite (for any 2 inside C) we a14 bf eoO-1 e@=1+ 55 f. at as in which the contour of integration can be taken to be the familiar contour made up from the real axis and a semicircle at infinity in the upper half-plane. The arc at infinity makes no contribution to the integral, Decause ¢ —1 — 0 suffidently fast, and one has, for any 2 with Sz >0We take 2=w +e, €>0, and deduce Lp? efw)=1 oe = Malla | a aunie Now we use oo ky fo gear |” [2 £0. 2 +i fz), whee PJ denotes the Cauchy prindpal value integral (recall sh i of 0 an ste = Timms sos fe f(@) de + [0,5 f(@)de), where the limit is wll-defined). Thus ele=1 ew)=14-dp [oe or separating real and imaginary pants Rew) =144P [Sas Sew) Lp [Baudet RaW =I ay, TE we ue also the previous result on ¢ which amounts to Rey(—w) = Rew) Seq(—w) = — Ber) there follows the Kramers-Kriinig relations Rea) =14 2p [EO ay Bp [Ray Ser(w) wat ‘which had an important influence on the devdopment of quantum mechanics, and which in their generalised {form continne to be of importance in studies of high energy physics. Because in the optical case they give dations between the real part of ¢., which determines dispersion, and the imaginary part which determines absorption, (ca. section (3.3)) these relations and their generalisations are called dispersion relations; they are a consequence of anabyticity (here of ¢.), Which in tum is intimately related to the amusadity of the physical proces studied.