CARL T. A. JOHNK Professor of Electrical Engincering University of Colorado, Boulder Engineering Electromagnetic Fields and Waves JOHN WILEY & SONS New York Chichester Brisbane Toronto SingaporeCopyright © 1988, by John Wiley & Sons, Inc. All rights reserved, Published simultaneously in Canada, Reproduction or translation of any part of this work beyond that permiteed by Sections 107 and 108 of the 1976 United States Copyright Act without the permission of the copyright ‘owner is unlawful, Requests for permission or further information should be addressed the Permissions Department, John Wiley & Sans. Library of Congress Cataloging-in-Publication Data ‘Johnk, Carl T.A, (Carl Theodore Adolf), 1919 Engineering electromagnetic fields and waves. Includes index. 1. Electric engineering. 2. Blectromagnetic fields. 3. Electromagnetic waves. I. Title TKI45J56 1988 6213 87-6159 ISBN 0-471-09879-5 Printed in the United States of America 10987654321 To Jeanette and the BoysCHAPTER 2 Vector Differential Relations and Maxwell’s Differential Relations in Free Space In this chapter is considered the development, in generalized orthogonal coordinates, of the gradient, divergence, and curl operators of vector analysis, with forms in the common. coordinate systems taken up in detail. The divergence theorem and the theo- rem of Stokes are used to derive the differential forms of Maxwell's divergence and curl equations in free space from their integral versions postulated in Chapter 1. ‘The appropriate manipulations of Maxwell’s time-varying differential equations are seen to fead to the wave equations in terms of the B and E fields, and the wavelike nature of their solutions is exemplified by considering in detail the field solutions of uniform plane waves in free space. A pursuit of these ideas requires some background in the differentiation of vector fields, to be discussed in the following section. 2-4 DIFFERENTIATION OF VECTOR FIELDS In many physical problems involving vector fields, a knowledge of their rates of change with respect to space, time, or perhaps some parameter is often of interest. This notion has already been introduced in Section 1-6 in connection with the position vector r. It is now considered in general for any differentiable vector field. If F(u) is a vector function of a single scalar variable z, its ordinary vector deri- vative with respect to is defined by the limit dE AR Ftv + Au) — Ft) —= h — 1 da Any Aw a au Gy 61 | ' } j62 VECTOR DIFFERENTIAL RELATIONS AND MAXWELL'S DIFFERENTIAL RELATIONS arenes ron F (u) Fu) Uo Au ~~ (A space variable) FIGURE 2-1. A vector function F in space, and its variation AF with respect to some variable 1. F (u + Au) Put aw) provided that the limit exists (i.e., the limit is single-valued and finite). As in the in- stance of the derivatives of the position vector r considered earlier, the vector inere- ment AF is not necessarily aligned with the vector F, implying that the direction of the vector F may change with the variable «. This circumstance is exemplified in Figure 2-1, in which the conventional triangle construction is used to define AF, the difference between F(u + Au) and F(u). The derivative dF/du defines a function, the derivative of which in turn defines a second-order derivative function d?F/du?, and so on. ‘The derivatives of the sum or product combinations of scalar and vector fune- tions are often of interest. For example, if f and F are respectively scalar and vector functions of the variable #, the derivative of their product is, from (2-1) a/F)_ Pe Anne ae it au (2-2) Note that this result resembles in form a similar rule of the scalar calculus fin which both functions are scalars). IfF is a function of more than one variable, say of a, u2, ws, f, its partial deriva- tive with respect to one of the variables (u,) is defined OF (ur, wa, ws.) Fly + Au, ta, us, t) — Ft, ta, ta, #) = li -3) Buy ano Aus 23) with similar expressions for the partial derivatives with respect to the remaininy variables, Successive partial differentiations yield functions such as 0°E/éu3, @E/éu, Guy. If F has continuous partial derivatives of at least the second order, it is permissible to differentiate it in either order; thus OF OF Buy Guz Guy Ou (24) The partial derivative of the sum or product combinations of scalar and vector functions sometimes is useful. In particular, one can use {2-3} to prove that the following expansions are valid .2-2 GRADIENT OF A SCALAR FUNCTION 63 auf) Fe ape eRe (2-5) 2-6) (2-7) iff is any scalar function and F and G are vector functions of several variables, among which ¢ denotes a typical variable. 2-2 GRADIENT OF A SCALAR FUNCTION ‘The space rate of change of a scalar field f(uy, #2, 3, ¢) is frequently of physical interest. For example, in the scalar temperature field T(u,, uz, #3, ¢) depicted in Figure 1-1 (a), one can surmise from graphical considerations that the maximum space rates of tem- perature change occur in directions normal to the constant temperature surfaces shown. Generally, the maximum space rate of change of a scalar function, including the vector direction in which the rate of change takes place, can be characterized by means of a vector differential operator known as the gradient of that scalar function, It is developed here. If, at any fixed time ¢, a single-valued, well-behaved scalar field f(y, ua, ts, 4) ig get equal to any conslant fy $0 that f(y, up, uz, £) = fo, a surface in space is described, as depicted by S, in Figure 2-2. A physical example of such a surface is any of the constant temperature surfaces of Figure 1-1(a). Another surface, $2, an infinitesimal distance from Sj, is described by letting flay + diy, uz + ditg, uy + dug) = fo + df, in which df is taken to mean a very small, constant, scalar amount. Suppose that two nearby points, P and P, are located a vector distance df apart on these two surfaces ‘Sq (defined by Feira) ap @ (on Sa) ‘dn = a, dn (@) eo) FIGURE 2-2, Two nearby surfaces f= fy and f = fy + df relative to a discussion of grad F(a} Points P and P' separated by df and on surfaces defined by f = fy and f=fo + a 10) Points P and P' on the same surface f = fo, to show that grad f and df are perpendicular,64 vector DIFFERENTIAL RELATIONS AND MAXWELL’S DIFFERENTIAL RELATIONS as in Figure 2-2(@}, recalling from (1-21) that one may express df = adé as dh = aydl, + ay df, + asdf, (2-8) df is the amount by which f changes in going from P to P from the first surface to the second, written as the total differential a= tay Ft Buy tt Bag Mt Buy af af = tt, - “a rca, +5 dt, (2-9) ‘The presence of the components of dé in (2-9) permits expressing df as the dot product - g g g w-[s itm tam |: (a, dé, + a dé, + a3 des) Calling the bracketed quantity the gradient of the function f, or simply grad f, as follows of Ce f grad f= aia t ag tae or Py Ete grad fear a Buz 7? hy dug (2-10) one may write the total differential df of (2-9) in the abbreviated form af = (grad f) dé (2-11) Two properties of grad f are deducible from (2-11): 1. That the vector function grad f defined by (2-10) is a vector perpendicular to any f = fo surface is appreciated if the points P and P’, separated by a distance dé, ave placed on the same surface as in Figure 2-2(6). Then the amount by which f changes in going from P to P’ is zero, but from (2-11), (grad f) - dé = 0, implying that grad f and dé are perpendicular vectors. Grad f is therefore a vector everywhere perpendicular to any surface on which f = constant. . Tf a displacement df from the point P is assigned a constant magnitude and a variable direction, then from (2-11) and the definition (1-34) of the dot product itis seen that df = lgrad f| dé cos 0, @ denoting the direction between the grad f and df. The magnitude of grad f is therefore df/(df cos 0), but from Figure -2(a}, dé cos 8 = dz, the shortest (perpendicular) distance from the point P on the surface 6, to the adjacent surface S, on which f = fy + df, whence Ns \grad f[= ¢ (2-12)22 GRADIENT OF A SCALAR FUNCTION 65 ‘The vector grad f therefore denotes both the magnitude and direction of the maximal space rate of change of f, at any point in a region. Note that the magnitude ofgrad_f can also be expressed in terms ofits orthogonal curvilinear components, given. in the definition (2-10) by bart pat pa ype leraa / I [GS (ci) +k) | a) The expressions for grad fin a specific orthogonal coordinate system are obtained from (2-10) on substituting into it the appropriate symbols for 1, and A; as discussed in Section 1-5. Thus, in the rectangular system : ee grad faa 4 aay tea (2-14a) in the circular cylindrical system a, Ly ar — bas 14) grad f= Bp 8p ag ae (2-14b) and in the spherical coordinate system oe lof 1 ¢ df =a, s grad f= a, 5+ ay ~ vat Sranbee (2-14) An integral property of grad /, of considerable importance in field theory, is that its line integral over any closed path ¢ in space is zero. Symbolically ¢ (grad f)-df =0 (2-15) holding for all well-behaved scalar functions f, and proved in the following manner. Consider (2-15) integrated over an open ‘path between the distinct endpoints Poluf, ue, v8) and Plats, ta, us) t (grad f)- dé (2-16) From (2-11) itis seen that (grad f) > dé denotes the total differential df, so that (2-16) becomes Sf wead py ae = [Par = J], = fle tay ma) ~ fl08, 18, 08) (2-17) or the difference of the values of the function f at the endpoints P and Pp. Thus, any path connecting Py and P will provide the same result, (2-17). Carrying out (2-16) over some path A from Pp to the point P and then back to Py once more over a different path B, the contributions of the two integrals would cancel exactly, making (2-15) the result. The integral property (2-15) of any vector field grad f is sometimes66 vector DIFFERENTIAL RELATIONS AND MAXWALL’S DIFFERENTIAL RELATIONS called the consereative property of that field, from the applications of integrals of that type to problems involving certain kinds of energy. Any field grad f is a conservative field 2-3 THE OPERATOR V (Del> ‘ Recall that the gradient of a scalar field f is expressed in rectangular coordinates by (2-1Ma) _.¥,.4, a grad faa, 5 bay sta, se [2-14a] The presence of the common function f in each term permits separating from this ex- pression a vector partial differential operator represented by the symbol V (pronounced del) as follows at é 3 (2-18) v de yt By to permit writing grad f in an alternative symbolism, Vf = F Ca oe grad f= Visas tay base (2-19) The notations grad f and V/ will henceforth be considered interchangeable. It may be noted that the operator V defined by (2-18) in the rectangular coor- dinate system can be defined in other coordinate systems as well, including the gen- cralized orthogonal curvilinear system. This is not done here because of its lengthy form and because it serves no particular need in connection with the objectives of this text, You may wish to consult other sources relative to extending (2-18) to other coordinate systems.! EXAMPLE 2-4. Suppose a scalar, time-independent temperature field in some region of a space is given by T(x, 9) = 200 + 1009 deg with x and y expressed in meters. Sketch a few isotherms (constant temperature surfaces) of this static thermal field and determine the gradient of T. The isotherms are obtained by setting T equal to specific constant temperature values. Thus, leiting T= 100° yields 100 = 200% + 100, the equation of the tilted plane 9 = —2 +1. This and other isothermic surfaces are shown in the accompanying figure. ‘The temperature gradient of 7(s, y) is given by (2-I4a) er aT ar a, = 100,” nay tA Gy Fae Ge = 2000, + 100," VI =grad T= ‘For example, sce M., Javid, and P.M. Brown, Field Analysis and Electromagnetics. New York: McGraw—EEll, 1968, p. 477.24 DIVERGENCE OF A VECTOR FUNCTION 67 2 (@) EXAMPLE 2-1, (a) Graph of T= constant, (6) Side view of (a). a vector everywhere perpendicular'to the isotherms, as noted in (6) of the figure. The x ‘and_y components of the temperature gradient denote space rate of change of temperature along these coordinate axes. From (2-13), the magnitude is [grad T| = 2007 + 100? = 293°/m denoting the maximal space rate of change of temperature at any pot, One may observe that heat will flow in the direction of maximal temperature decrease; that is, along lines perpendicular to the isotherms and thus in a direction opposite to that of the vector grad T at any point 2-4 DIVERGENCE OF A VECTOR FUNCTION The flux representation of vector fields was described in Section 1-9. Ifa vector field F is representable by a continuous system of unbroken flux lines in a volume region as shown, for example, in Figure 2-3(a), the region is said to be source free; or equivalently, the ficld F is said to be dizergenceless. (The divergence of F is zero.) On the other hand, Arbitrary closed 7) ( FIGURE 2-3. Goncerning the divergence of flux fields, (2) A vector field F in 2 source-free region. As many flux tines enter S as leave it. (}) A vector field F in a region containing sources {S.possessing net outgoing flux). (¢) The meaning of div F: net outward flux per unit volume as Avs.GB VecTOR pirreRrNTIAL RELATIONS AND MAXWRLLE’S DIFFERENTIAL RELATIONS if the flux plot of F consists of flux lines that are broken or discontinuous, as depicted in Figure 2-3(6), the region contains sources of the field flux: the field F is then said to have a nonzero divergence in that region. The characterization of the divergence of a vector field on a mathematical basis is described here. ‘The divergence of a vector field F, abbreviated div F, is defined as the limit of the nct outward flux of F, §s F + ds, per unit volume, as the volume Av enclosed by the surface 5 tends toward zero. Symbolically, div F = lim fra flux Hines/in3 (2-20) Avo Thus, as the closed surface Sis made very small, as depicted in Figure 2-9(c), the limit- ing, net outward flux per unit volume in the neighborhood of the point P defines the divergence of the vector field F there. The shape of Sis immaterial in this limit, as long as the dimensions of Av tend toward zero together. ‘The definition (2-20) leads to partial differential expressions for div F in the various coordinate systems, For example, in generalized orthogonal coordinates, div F is shown to become f wo dvr! | catats + AUF ghyhs) 4 eee Iyhzh| bay Bu Gas (2-21) ‘The derivation of the differential expression (2-21) for div F in generalized ortho- gonal coordinates proceeds from the definition (2-20). Express the function F in terms of its generalized components as follows = Pty. ty ts) = ay (up, tas ta, 6) + mal y(uy, up, ug, t) + aghy(uy, up, ws, 6) (2-22) ‘The definition (2-20) requires that the net effiux of F be found over the closed surface S bounding any limiting volume Av, which from (1-11) or (1-18) is expressed Av = Ad, Ady Ads = hyhghy Aus Auy Aug (2-23) ‘The net, outward flux of F is that emanating from the six sides of Av, designated by As,, Asi, and so forth, in Figure 2-4(a). The contribution A, entering element As, is just FAs, = (ayF,) + Asy, or FAs, = F+(—a, At, Ads) = —F, Ad Af (2-24) = —F highs Auy Auy = —Ay, (2-25) the negative sign being the consequence of assuming a positively directed F, component over the outward As, = ~a, A¢, A¢3; that is, the flux Ay, enters As,. In the limit, as the separation A¢, between As; and As, becomes sufficiently small, the flux AW’, leaving As, in Figure 2-4(b) differs from Ay, entering As, by an amount given by the second24 DIVERGENCE OF A VEGTOR FUNCTION 69 Flux through As: = —Pibbaly = - Ay {tua) i Flue through asi: am aida + AD uy @ w FIGURE 24, A volume-clement Az in the generalized orthogonal coordinate system used in. the development of the partial differential expression for div F. (a) A volume-element Av and components of F in the neighborhood of P(u,, v2, xs). () Flux contributions entering and leaving opposite surfaces of Av. The remaining four sides are similarly treated. term of the Taylor’s expansion of Ay, about the point P; that is, oa AW (uy + Aug, ua, us) = Ady + Au, = FMA + lx (Fj At ads) an 7 = Fy Aly My + [4 Fut) Ati, Au, Atty (2-26) 7 Itis permissible to remove Au, and Au, from the quantity affected by the 6/du, operator in the foregoing because each is independent of w;, in view of the orthogonal coor- dinate system being used. The net outgoing flux emerging from the sides As, and Asi of Figure 2-4(6) is thus the sum of (2-24) and (2-26) é AW, — Ad, = 5 (Pitas) At Ait Any (2-27a) ‘The two remaining pairs of surface elements As}, As, and As,,, As) similarly con- wibute outgoing flux in the amounts A, — Ad = 2 (Fyhsha) Ay Any Aug (2-27b} 2 a nA ~ As = 5 (Pola) Any Auz As (2-27¢) 3 scen to be obtainable from the symmetry and the cyclic permutation of the subscripts of (2-27a). Finally, putting (2-27a, b, c) into the numerator of (2-20) obtains the result70 Vector DIFFERENTIAL RELATIONS AND MAXWELL’S DIFFERENTIAL RELATIONS anticipated in (2-21) iv Pe lim ES _ AV — Av, + Atha — Av, + AWS — Avy OPS mo fim Tihghs Au, Aug Aug whence divF = Mish) (2.20) In rectangular coordinates, div is found from (2-28) by setting A = hy = hy = 1 and iy = 44 a =, ty =z a, OF, aF, iv B= Se Oy 2.99, div B= Se bog +52 Rectangular (2-29a) Whereas in the circular cylindrical and the spherical coordinate systems, the expressions become , 1a 1 éFy @F, . eos div Fo tek) + oot ae Gircular cylindrical (2.29) . lay, Loa. div P= eR) + sin 8) 1 éF, . + ae Re Spherical (2-29¢} ‘The form (2-298) of div F in the rectangular system is the basis for another nota- co eins the del operator (V) defined by (2-18). On taking the dot product of V with F in the rectangular system of coordinates, one finds é é é VeF= (2+ ayes) (@,F, +a,F, tak) OP, aF, ar et B TR (2-30) or precisely (2-292). This is the basis for the equivalent symbolisins dvF=V-F (2:31) The notations diy F and V+ ¥F will be considered interchangeable regardless of which coordinate system is used, even though the symbol V has for our purposes been defined only in the rectangular system.Gs ayky on om ( = ak 2a =—). 1 1 - ~ 24 DIVERGENCE OF A VECEOR FUNCTION 71 : 1 1 i Test closed surface $ EXAMPLE 2-2 EXAMPLE 2. .. Sketch flux plots for each of the following vector fields, and find the divergence of each: (a) F = a,k, G=a,ky, H= a,x; (b) Ja a,K, L = a,(X/p). {a) Applying (2-29a) to the functions F, G, and H in the rectangular system obtains =O Mg vey civ — 8) vP a oe és K ‘Their flax plots are shown in (2) through (c}. Inspection reveals why a zero value of divergence is obtained for the fields F and G; a test closed surface § placed any- where in the region will have zero net flux emanating from it, The nonzero div H, on the other hand, is evident from its flux plot because of the discontinuous flux lines, here required to possess an increasing density with x, yielding a net nonzero outgoing flux emerging from the typical closed surface $ shown. (8) From (2-29) 1¢@ die J= 550K) the flux plots of which are illustrated looking along the z-axis of the circular ey- lindical system in (d}. The divergenceless character of L is evident from its I/p dependence, which, in this cylindrical system, provides an uninterrupted system of72. vecror DIFFERENTIAL RELATIONS AND MAXWELL'S DIFFERENTIAL RELATIONS i outgoing flux lines. The radially directed field J, having a constant flux density of magnitude , on the other hand, clearly must pick up additional flux lines with an increase in p. It is therefore required to possess a divergence. EXAMPLE 2-3. Find the divergence of the E field produced by the uniformly charged cloud of Figure 1-13() at any location P(r) both inside and exterior to the cloud. ‘The field E(1) outside the cloud {r > rq) is given by (1-59). Tts divergence (2-29c) in spherical coordinates is 3 (PB) = sal? oe] =0 1>% (2-82) ‘This null result signifies a flux plot in the region r > ro consisting of unbroken lines, as noted in Figure 1-16(6). All inverse 7? radial fields behave this way Inside the charged cloud (r < rg), the E field (1-60) being proportional to r has the divergence L ave= 5 (#fe)—% r<% (2-33) ? © a nonzero, constant result, proportional to the density g, of the cloud. Note that bringing &9 inside the divergence operator puts (2-38) into the form div (€9E) = p,C/m* or de) Feds = Tim S (div F) do = fi (div B) ao (2-37) Ss Avid i= 1 v or just (2-84), known as the divergence theorem. If the limiting process yielding (2-37) is to be valid, ic is necessary that F, together with its first derivatives, be continuous in and on V. If F and its divergence V+ F are not continuous, then the regions in V or on S possessing such discontinuities or possible singularities must be excluded by constructing closed surfaces about them, as typified in Figure 2-5(6). Note that the volume V of that figure is bounded by the multiple surface § = Sy +8, +3, with S; and $3 constructed to exclude discontinuities or singularities inside them. The normal unit vectors a,, identified with each vector sur- face clement ds =a, ds on S,, 53, and 55, are assumed outward unit vectors pointing away from the interior volume F. The following examples illustrate the foregoing remarks concerning the diver- gence theorem. EXAMPLE 2-4, Suppose the one-dimensional field Hix) = a,Kx of Example 2-2(a) exists in a region. Illustrate the validity of the divergence theorem (2-34) by evaluating its volume and surface integrals inside and on the rectangular parallelepiped bounded by the co- ordinate surfaces = 1,x=4,)=2,9 = —2,2=0, and z= 3, for the given H.74 VECTOR DIFFERENTIAL RELATIONS AND MAXWELL’S DIFFERENTIAL RELATIONS \(z) Sa — agdy dz 07 Sa o) @) EXAMPLE 2-4 Since div H = K, the volume integral of (2-34) becomes (div 31) de = Ie 3p. ar a) Evaluating the surface integral requires summing the integrals of H » ds over the six sides of the parallelepiped. Because H is x-directed, however, H+ ds is zero over four of these sides, the surface integral reducing to the same result as (1) Gyaeas= [0 [jean acdaet [oP ult (abe) = 48K — 12K = 36K (2) : EMAMPLE 2-5. Given the p-dependent field: E = a,K/p", with Ka constant, illustrate the Ss (End cap) (Detail of thin tube sed to exclude ssinguiarity) EXAMPLE 25 validity of the divergence theorem by evaluating both integrals of (2-34) within and on aright circular cylinder of length Z, radius R, and centered about the z-axis as shown,24 DIVERGENCE OF A VECTOR FUNCTION 75 Since E has a singularity at p = 0, a thin, tubularsurface 6, of radius a is constructed as shown, to exclude the singularity from the integration region, making § — S, + S$, Ss + S,. The divergence of E, by use of (2-29), is ye yielding the following volume integral Jew Pe [ieee = antic — oN) ay Jena pil? With E p directed, the surface integral of (2-34) reduces to contributions from only 3, and $2 (the endeaps yielding zero outward flux), whence * K * K fBras= f(t, (2) taRep ae) + [fr {=.ct): (—a,edd dz) = KL (RY? — aX) @) agreeing with the result {1}. [Note: Bach answer has the limit 2nKLR"? as a > 0.] The usefulness of the divergence theorem embraces more generally the inter- change of volume for closed-surface integrals required for establishing several theorems of electromagnetic theory. An example occurs in Poynting’s theorem of electromagnetic power considered later in Chapter 7 B. Maxweil’s Divergence Relations for Electric and Magnetic Fields in Free Space The definition (2-20) of the divergence of a vector field serves as a basis for de- riving the differential, or point, forms of two of Maxwell's equations from their cor- responding integral forms (1-53) and (1-54) for free space $ (608) ds =f) py [1-53] & ‘These laws apply to closed surfaces of arbitrary shape and size. If § is the surface bounding any smail volume element Az, dividing (1-53) by Av yields 45 (6B) “ds _ fy p.do Av “Re ds =0 Wb [1-54] (2-38) ‘The limit of the left side, as Av becomes sufficiently small, is div (€,E) from the definition (2-20). The right side denotes the ratio of the free charge Ag inside Ac to Av itself its limit is p,. As Av — 0, therefore, (2-38) becomes div (€gE) = p, C/m3 (2-39)76 vector DIFFERENTIAL RELATIONS AND MAXWELL'S DIFFERENTIAL RELATIONS the diffzrential form of Maxwell’s integral expression (1-53). Note that expressing (2-39) in rectangular coordinates using (2-29a) yields the partial differential equation ah, ah, OE, _ py =a 1-40) aoe & (2-40) It is evident that the divergence of (gE) at any point in a region is precisely p,, the volume density of electric charge there, implying that the flux sources of E fields are electric charges. Equivalently, if electric field lines terminate abruptly, their termini must be electric charges. By similar procedure applying (1-54), one obtains the following partial differen- tial equation in terms of B div B = 0 Whjm? (2-41) implying that B fields are always divergenceless and therefore source free. The flux plot of any B field must, therefore, invariably consist of closed lines: free magnetic charges are thus nonexistent in the physical world. A divergenceless field is also called a sulmoidal field; magnetic fields are always solenoidal. EXAMPLE 2-6. Suppose that Maxwell’s differential equation (2-39), instead of its integral form (1-53), had been postulated, Execute the reverse of the process just described, deriv- ing (1-53) from (2-39) by integrating the latter over an arbitrary volume V and applying the divergence theorem. Integrating (2-39) over an arbitrary volume V’ yields f div (€gE) és = f Pode Assume that E is well-behaved in the region in question. From a use of (2-34), the left side can be replaced by the equivalent closed-surface integrai § {€9B) * ds, and (1-53) foliows & (€0E) * ds = a pedo (1-33] 2-5 CURL OF A VECTOR FIELD From (2-15) ir is established that the line integral of (grad _f) +d around any closed path is always zero. Many vector functions do not exhibit this conservative property; ‘ physical example is the magnetic B field obeying Ampére’s circuital law (1-56). For example, in the steady current system of Figure I-19, the line integral of B- df taken about a circular path enclosing all or part of the wire, a nonzero current result is anticipated. Nonconservative fields such as these are said to possess a circulation about closed paths of integration. Whenever the closed-line integral of a field is taken about a small (vanishing) closed path and the result is expressed as a ratio to the small area enclosed, that circulation per unit area can be expressed as a vector known as the curl of the field in the neighborhood of a point. It follows that a conservative field has a zero value of curl everywhere; it is also called an irrotational field.2.5 CURL OF A VEGTOR FIELD 77 Paddle wheel FIGURE 2-6. A yclocity feld in a fluid, with an interpretation of its curl from the rotation of a small paddle wheel. Historically, the concept of curl comes from a mathematical model of effects occurring in hydrodynamics. The early work of Helmholtz in the vortex motion of fluid fields lec ultimately to the mathematical postulates by Maxwell of Faraday’s con- ceptions of the electric fields induced by time-varying magnetic fields. A connection between curl and fluid phenomena can be established by supposing a small paddle wheel to be immersed in a stream of water, its velocity field being represented by the flux map shown in Figure 2-6. Let the paddle wheel be oriented as at A in the figure. The effect of the greater fluid velocity on one side than on the other will cause the wheel to rotate—-clockwise, in the example shown. In this example, the velocity ficld v is said to have a vector curl directed into the paper along the axis of the paddle wheel, a sense determined by the thumb of the right hand if the fingers point in the direction of the rotation; the vector curl of» has a negative z direction at A. Similarly, physically rotating the paddle wheel axis at right angles as at B in the figure provides a way to determine the x component of the vector curl of », symbolized [eurl 2],. In rectangular coordinates, the tolal vector curt of » is the vector sum curl » = a,[eurl »], + a,[eurl oJ, + a,[eurl v], Generally, the curl? of a vector field F(u;, 12, us, 0), denoted eurl F, is expressed as the vector sum of three orthogonal components, as follows curl F = a, {curl F], + a,[eurl F], +a;[curl F]; (2-42) Each component is defined as a line integral of F - df about a shrinking closed line on a per-unit-area basis with the a; component defined Fed a,[curl F], =a, lim SF de 2-43 wy Ay (48) The vanishing surface bounded by the closed line ¢ shown in Figure 2-7 is As,, with the direction of integration around ¢ assumed to be governed by the right-hand rule. In Europcan texts, eux F is written rot F, and is read rotation of F. “The integration seuse coincides with the direction in which the fingers of the right hand point if the thumb points in the direction of ay.: 78 VECTOR DIFFERENTIAL RELATIONS AND MAXWELL’S DIPFERENTIAI. RELATIONS Integration ‘ sense ey FIGURE 2-7. A closed line ¢ bounding the vanishing area As,, used in defining the a, component of curl F at P. Similar definitions apply to the other two components, so the total valuc of curl F at a point is expressed . at _ Bed a? cut a ta ale Ag Tan By OD A differential expression for eurl Fin gencralized coordinates is found from (2-44) by a procedure resembling that used in finding the differential expression for div F in Section 2.4, The shape of each closed line ¢ used in the limits of (2-44) is of no con- sequence, as long as the dimensions of As inside ¢ tend toward zero together. Thus, in finding the a; component of curl F, ¢ is deformed into the curvilinear rectangle of Figure 2-8() with edges A¢) and A¢y. The surface bounded by ¢ is As; = a, Ad, Ady = ayhghs Au, Aus, the only components of F contributing to the line integral in the numerator of (2-43) being F, and F;. Thus, along the bottom edge A¢2, the contribution to §, F- dé becomes Aw = F, At, (2-45) in which Aw, denotes that contribution. Along the top edge, F; changes an incremental amount, but in general so does the length increment, A¢, because of the curvilinear coordinate system. The line-integral contribution along the top edge is found from a Taylor's expansion of Aw’, about P. The first two terms are sufficient if Aus is suitably Integration. -w Aus = Fd} ob) FIGURE 2-8. Relative to eu#l Fin generalized orthogonal coordinates. {a} ‘Phe components of F at typical point P, (#) Construction of a path £ relative to the a, component of curt F.25 CURL OF A VECTOR FIELD 79 small; thus Aw (uj, ug, uz + Aus, Ou Gus : (2-46) the negative sign resulting from integrating in the sense of decreasing uz along the top edge -[ae, 4 SAen) a] = -[s At, + Palo) A | “Similarly, the contribution along the left edge Ady in Figure 2-8(b) is Aw; = —F, Mts (2-47) whereas along the right edge, it is O(F Ads) Au = Fy Aes + 5 Any, (2-48) ‘The substitution of (2-45) through (2-48) into the definition (2-43) obtains for the ay component of curl F a,[eurt F), GR dé ay + Aw + Avog + A's =e on a dm, AG AC, UA fr, At, [* At AEN a, | = Fy My + [rs Ms +E an} z 1 0. So if G were a fiuid velocity field with a paddle wheel immersed in it as in Figure 2-6, a clockwise rotation looking along the negative z direction would result, agreeing with the direction of eurl G. EXAMPLE 2-8, Find the curl of the B fields both inside and outside the long, straight wire carrying the steady current / shown in Figure 1-19. 28 CURL OFA vecToR Frey 81 ‘The B field is given by (1-64), a @-directed function of p. The curl of B, obtained from (2-51), yields inside the wire (p < a) a, a: Ma Pe * 2 a al_ wat slp I wBaj— FS) OY stole) lp Ob se [onl aaa? e Hole 0 pf Mele e [4] a result proportional to the current density J, = I/na* in the wire. This special case demonstrates the validity of a Maxwell's differential relation to be developed in Section 2-5B. You may further show from (2-54) that eux B ouiside the wire is zero, in view of the inverse p dependence of B there. A, Theorem of Stokes Tf Fly, ta, ts, #) is well-behaved in some region, then the integral identity LW x8) asf rae (2-56) holds for every closed line ¢ in the region, if $ is a surface bounded by ¢. This is called the theorem of Stokes. A heuristic proof follows along lines resembling the proof of the divergence theorem. Suppose the arbitrary S is subdivided into a large number n of surface-clements, typical of which is As, bounded by ¢;, as in Figure 2-9(a). The line integral of F + dé around ¢; is inferred from the definition (2-48) of the component of the eurl F in the direction of As; to be Fae = [eurl F] + As; (2-57) ds = ands 4 t 8 Qe Integeatiof sense of & FB ede fahthtty (6) FIGURE 2-9. Relative to Stokes’s theorem. (a) Showing a typical interior surfaces element As, bounded by ¢3. (b) Closed lines ¢ and 4 constructed to eliminate ais. continnities from 5.82 VECTOR DIFFERENTIAL RELATIONS AND MAXWELL'S DIFFERENTIAL RELATIONS for As; sufficiently small, If the left sicle of (2-57) is summed over all closed contours Z, on the surface § of Figure 2-9(a}, the common edges of adjacent elements arc traversed twice and in opposite directions to cause the integrations about ¢; to cancel everywhere on S except on its outer boundary ¢. Summing the left side of (2-57) over the 7 intetior elements As; therefore obtains S|: p wea] = Gea (2-58) and equating to the right side of (2-57) summed over the same elements yields the result, asm approaches infinity LF dé = tim ¥ [(curl F) + As) = Ji (ewd F) ds (2-59) AsO 1 which is Stokes’s theorem (2-56). ‘As with the divergence theorem, it is necessary in (2-56) that F together with its first derivatives be continuous. Ifnot, the discontinuities or singularities are excluded. by constructing closed lines about them as in Figure 2-9(4), causing § to be bounded by the closed line = 4, +4 +¢y. The connective strips, of vanishing widths as shown, are however, traversed twice so their integral contributions cancel. The positive sense of ds should as usual agree with the integration sense around ¢ according to the right-hand rule. EXAMPLE 2-9. Given the vector ficld Fong, 2) = aby + ayy? + apse a) illustrate the validity of Stokes’s theorem by evaluating (2-56) over the open surface S defined by the five sides of a cube measuring ] m on a side and about the closed line ¢ bounding S as shown. Positive side ofS Positive integration sense (x) +l tht h) (SS) 4 +85) {a) (a) EXAMPLE 2-9. {¢) Line clements on ¢, (6) Surface elements on S,25 CURL OF A VECTOR Fietp 83. The ne integral is evaluated first. The right side of (2-56) applied to £ becomes, making use of figure (a) Gem [Rade Radek PR add [Rade =o4 ft a a $ (2) ‘The surface integral of (2-56) is found next. From (2-52). a, ay a, 6 8 @ oul P=) 5 [tac tai ase 8) Sox 9 oz ‘whence the surface integral of (2-56) evaluate over Si, .. 8 yields, using figure (6), top Siren asa 0 fi eclodet~ fh fh, [le=o daz opt wpe [a Boeclendea + Ef, ileor dee — Jeg fig Burpeodede = 3 a which agrees with (2). EXAMPLE 2-49. Given the vector ficld F(9) =a, cot a in which & isa constant, illustrate the validity of Stokes's theorem by evaluating (2-55) for the hemispherical surface § with a radius a, hounded by the closed line / at 8 = 90", 7 — a as shown. kere isa singularity in F on Sat@ = 0°; icmust be excluded to assure the validity of Stokes’s theorem on the given surface. To accomplish this, a smail eircle ¢, at 8 = 0, and 1 = ais constructed as in (6). If ds is assumed positive outward on S, then the sense of the line integration is as noted, the integrals cancelling along Z, and /, of the connective strip Singularity in F rs Integration sense (a) (B) EXAMPLE 2-10. (a) Open hemispheric surface 8. (2) Exclusion of the singular point84 —vecTOR DIFFERENTIAL RELATIONS AND MAXWELI'S DIFFERENTIAL RELATIONS as its width vanishes. The line integral around ¢ = ¢, + ¢3 thus becomes Be des [2% Fe agrsin 04d ° = Ka cos 0,{9]5** = —2naK cos y+ SLE F ayrsin 040] ‘va converging to —2naK’ as 0, > 0. The surface integral is evaluated using ay ay ay rsind rsind r curlF=| 2 a ee & 06 66 0 0 (rsin 8) K cot | whence Jf (cor F) as eo ie. (Ka sin 0.d0.db} = —2naK cos 0, (3) which agrees with (2). You might consider how the results would have compared had one ignored the singularity B. Maxwell's Curl Relations for Electric and Magnetic Fields in Free Space In Section 2-4B, the divergence of a vector function was put to use in deriving the differential Maxwell equations (2-39) and (2-41} from their integral versions (1-53) and (1-54). The definition of the curl may similarly be used to obtain the differential forms of the remaining equations (1-55) and (1-56). Because the latter are correct for closed lines of arbitrary shapes and sizes, one may choose ¢ in the form of any small closed path bounding a, As, in the vicinity of any point, as in Figure 2-7. Taking the ratio of (1-35) to As, yields, with the assignment of the vector sense a, to each side, a GE de ~Fha Be ay ay i 1 (2-60) \ From (2-43), the left side, as As, + 0, becomes aj[eurl E],. The right side denotes the time rate of decrease of the ratio of the magnetic flux Ai, to As,, but this is just the component By at the point P. The limit of (2-60) therefore reduces to a,{eurl E], = fab) (2-61) relating the a, component of eurl E to the time rate of decrease of the a, component of the magnetic flux density B at any point.* The choice of the direction assigned by “The partial differentiation symbol 4/ appearing in (2-61) replaces the total differentiation a/dt in (2-60), in view of the fact that the field B is a function of space ax well as of time, whereas the volume integral in (2-60) is a fonction of ¢ only, for a fixed As,26 SUMMARY OF MAXWELL'S EQUATIONS: COMPLEX, TIME-HARMONIC FORMS 85, a1 is arbitrary, implying that two similar results aligned with the directions of the unit vectors a, and a3 and independent of (2-61) are also vali Combining these vectorially thus obtains the total curl of E at the point a a@,feurl E], + a,[eurl E], + a;[eurl E], TZ [a,B, + a,B, + a3B3) Making use of the notation of (2-42) yields the more compact form ee BR = OB i? VxE= ay Vim (2-62) a the differential form of Faraday’s law (1-55), Equation (2-62) states that the curl of She field E at any position is precisely the time rate of decroxse of sve ficld B there. “This implies that the presence of a time-varying magnetic feld Bin ¢ region is respon- sible for an induced time-varying E in that region, such that (2-62) is everywhere satisfied, A procedure similar to that used to derive (2-62) is applicable to the Maxwell relation (1-56), yielding the differential equation Vx— i {2-63} lo Tt states that the curl of B/ilg at any point in a region is the sum of the electric current density J and the displacement current density @(€9E)/ét at that Point. If the electric and magnetic fields in free space are static, the operator a/ét appearing in (2-62) and (2-63) should be set to zero, ‘This restriction provides the following curl relations for time-static fields VxE=0 (2-64) B Curl relations for static E, B fields Vx— Ho (2-65) Equation (2-64) states that any static F field is irrotational (conservative), whereas (2-65) specifies that the curl of a static B field at every point in space is proportional to the current density J there, 2-6 SUMMARY OF MAXWELL'S EQUATIONS: COMPLEX, TIME-HARMONIC FORMS One may recall that in Sections 2-4B and 2-5 the differential Maxwell equations for Iiee space were obtained from their integral forms, (1-53) through (1-56). These are collected for reference in Table 2-1, columns Land TIL "The integral Maxwell equationsloser {ee-a on leo-al on ou on 2 s 5 4 spe s 4 2300 + f=— 09) 2 4 p=—x sp. go" of ter -L°[ = 27?- ep af eee £f=7-—@ gore lage a a) aay pg {otter Lf Py prof p LP =e 2 5 4 spe 4 - Ls Bie wq'{ of =27- oy) a2 = 9- (eeu) gol- = aX A teow gp axa ap alo was [oct] 9? ayy wah few) -0= 9-4 lye o= aa o=-a'f bes) o= yap focal og = (go) [oe-a) °4 = (a) «A apig “f =or-32 [gor apta “f= na? SINOWRIVHTIINL INSONSERC-SNL SINOWRIVE-IL Xa1ANOD JNSONIECINI ‘adNOO ‘A wuaniof pruose gh suanof aSorey Bonds GO} U} SUOONDS S[ONKOW JO SUES ‘DIUOUUDY-SLUY, XE{ALOS PUD yuepuedep-oul, 2 F18VL2.6 SUMMARY OF MAXWELL'S EQUATIONS: COMPLEX, TIME-HARMONIC FORMS 87 were seen in Section 1-1] to be well suited for finding the field solutions of static charge or current distributions possessing simple symmetries, though methods relying on symmetry are unfortunately limited to a few isolated problems. The differential Maxwell equations usually offer amuch broader class of solutions; obtaining a number of these solutions wil} constitute the task of much of the remaining text. ‘Also of importance are the sinusoidal steady state, or time-harmonic solutions of Maxwell's equations. Tie-harmonic fields E and B are generated whenever their charge and current sources have densities varying sinusoidally in time. Assuming the Sinusoidal sources to have been active long enough that the transient field components have decayed to negligible levels permits the further assumption that E and B have reached a sinusoidal steady state. Then E and B will vary according to the factors cos (@t + 4,) and cos (ct + @,), in which 0, and 0, denote respective phases and « is the angular frequency. The alternative and equivalent formulation is achieved if the fields are assumed to vary according 10 the complex exponential factor o/*. This assumption leads to a reduction of the field functions of space and time to functions of space only, as observed in the following. ‘The field quantities in the real-time forms of Maxwell’s equations presented in columns I and ITI of Table 2-1 are symbolized BSE, 0, %3,) BS Blast, ts, 4) Tals, 34) Po = Polttay Uy Ua, 8) (2-66) “The linearity of the Maxwell relations guarantees that sinusoidal time variations of charge and current sources produce E aad B fields that in the steady state are also Sinusoidal, Then one may replace the functions (2-66) of space and time with products of complex functions of space only, multiplied by the complex factor ef as follows Eu, to, ug, 0) is replaced with B(x, uz, us)" Blu,, up, 13, 0 is replaced with Buy, ua, u3)e"" Jeu, My, wy, fis replaced with Flin, #2, us)e" Peltz, tay tty, #) is replaced with p,(uy, up. us)" (2-67) If the complex. vectors £, B, and J are written in terms of the generalized co- ordinate system as follows, that is, Blu, ta, 1g) = arb, + ark, + mab (2-68) then on inserting (2-67) into the Maxwell equations of column IIT, Table 2-1, one obtains V> (eB) = pel Vx & £ Bem a B , sion 8. eel V- (Bei) = 0 vx G =) = Jo + = (eoKel)——_(2-69) Ho ot “The partial-derivative operators V + and V x of (2-69) affect only the space-dependent fimetions Blu, ug, ty) and Blin, tz, tg), whereas 0/2¢ operates only on the eM factors88 vector DIFFERENTIAL RELATIONS AND MAXWELL’S DIFFERENTIAL RELATIONS common to all the fields. Equations (2-69) therefore yield, after cancelling the factars, V* (colt) = A, Cm} (2-70) =0 Wb/m? (2-71) Vx B= ~joB Vim? (2-72) Bo. ~ Vx —=J4+joe& Afm? (2-73) . Ho These are the desired complex, time-harmonic Maxwell equations for free space. They represent a simplification of the real-time forms in that the time variable ¢ has been eliminated. On finding the complex solutions E(w, u;, u3) and B(u;,, i, uy) that satisfy (2-70) through (2-73), the sinusoidal time dependence can be restored by multiplying cach space-dependent complex solution E and B by e!®* and taking the real part of the result as follows Elis ta, ta. 0) = Re [B(uy, ua, ug)e] Bly, ty, Wy. 8) = Re (Blu, up, use) (2-74) Considerable use is to be made of (2-70) through (2-74} in subsequent discussions of the time-harmonic solutions. One can show that a similar procedure using the replacements (2-67) leads to a complex, time-harmonic set of the integra? forms of Maxwell's equations in free space. A comparison with their time-dependent versions is provided in Table 2-1. Applications of the complex time-harmonic forms (2-70) through (2-73) to ele- mentary wave solutions in free space are considered in Section 2-10. A preliminary discussion of the Laplacian operator and a development of the so-called wave equations are desirable prerequisites to finding such solutions. These are discussed next, 2-7 LAPLACIAN AND CURL CURL OPERATORS ‘The gradient of a scalar field was seen in Section 2-2 to yield a vector field. Moreover, the divergence of the vector function grad f, denoted symbolically by V+ (Vf), is by the definition (2-20) a sealar measure of the flux source-per-unit-volume condition of Vf at every point in a region. The expansions (2-10) and (2-28) for Vf and its di- vergence can be combined to obtain V - (¥/) ina desired coordinate system, a result to be found usefisl for obtaining both time-varying and time-static field solutions. Thus, in generalized coordinates, the gradient of f is expressed by (2-10) waa lo 1 oF Lae tng Guy hy bug * Fy Guy (2-75)27 LAPLACIAN AND CURL OPERATORS 8G To find the divergence of Vf, the components of (2-75) become the elements Fy, Fy, and F of (2-28), obtaining yp ee (tats F hghy OF fishy Of v =a |an (St Bu, +e Ty Buy, +e ty Bu (2-76) This scalar result has a particularly simple form in rectangular coordinates, becoming . a=yy ot oF VW sv “Fa eS Q-77) The definitions of the dot product and of ¥ are seen to permit the following operator notations a VeVeG+5 zy (2-78) in which the notation V?, called the Laplacian operator, is equivalent to V-(¥) = V-¥( } =div (grad). From (2-76), the Laplacian operator in generalized coordi- nates is, therefore 1 é Ay 0 a ty 8 é f{hyhy 2 vaveve alts les, yh halla male (® du,)* 20; hg Bug) * Gag hy Bas, (2-79) yielding in the circular cylindrical system vravey al (ot) EE a (2-80) while in spherical coordinates yhe(ed)s 8 (aot) he VP al’ a) + sind 20 (8 859) tama ag? (BD90 VECTOR DIFFERENTIAL RELATIONS AND MAXWELL’S DIFFERENTIAL RELATIONS ‘The Laplacian operator (2-79) is also applicable to a vector field F(a, 2, #3. 0, the result of which is shown to be useful in the expansion of eurl (eurl F). Apply (2-79) to define VF, the Laplacian of a vector field, as follows vy 6 hy bug, 1 a (hzhs 6 oft 2 ye fighighs | Ouy hy Guy)” Bug hy Ourg) ” Buy x (aFy + agFy + a3F5} (2-82) WE ‘The term-by-term expansion of the latter can be tedious, since in general ay, a2, and a, are not constant unit vectors in a region; that is, their directions depend on a, #2, and ug. In rectangular coordinates, however, (2-82) yields the relatively simple result (since a,, a,, a, are constant unit vectors) WF =a,V2F, + a,V2F, + a,V2F, (2-83) in which the components VF, and so on, are specified by (2-77). No corresponding simplicity occurs in other coordinate systems because of the spatial dependence of the unit vectors already noted. For example, if the space partial derivatives of the unit vectors are properly accounted for, as in Example 1-1 of Section 1-6, one can show from (2-82) that ¥°F in the circular cylindrical system becomes iF, &, viv = ay] V8, Sie | tas] Vt SP Fe [+ ave, (2-84) a result decidedly not of the form of (2-23) with F), Fy, F, merely taking the places of Fy, Fy, F.. Sull another vector result, the curt of the vector curl F, designated V x (V x F), | is of importance. ‘The function V x F provides the three components given in (2-50); then performing another curl operation yields a [ty (AbaFa) albus) xB eT stasis ( éu Gag af hy (AmF) Oaks) ~ Gus hh — Ouy Athslis) _ OUdals) hs ms i Buy Quy 8lligFy) — 2th) -2 * ie 7 = {2 @hyFs) _ AbgFs) +o Ba a jus, ( OlhsFs) _ OligF) -£[A Buz us MN} (2-85)2.7 LAPLACIAN AND CURL OPERATORS 91 Because of its complexity, this result is examined only in rectangular coordinates, be- coming _ $2 om aR) 0 (ah, oF, veiverinad (5: ea Se) af (He _ OF) _ 0 (8%, _ Oe EG) Ne 0 (0 OF) _ 6 (0F, oF, tdalea)-5G- Zh A comparison of the latter with the vector ¥(V-B) is now made. In rectangular codrdinates, using (2-10} and (2-28) obtains OF, OF, OF, a (oF, oF, VV) ma Ge tay ae targa + ~[3(F +5, ) @ (oF, OF, 6 (aF, , ar, sols) lele-F)] en and adding and subtracting six properly chosen terms puts (2-87) into the following form Vive) = Oe ae tae Or Ok, PR be Pee (“8 OF, 7B) +a, + (Se Or, oF) a2 oP ae af 2 (PR OR) BR, 2 “ale a) a 32 a] 2 (% _@F,_ @F, Lay ae a2 Oe gal 2 (Ce, ah) _ 2h _@F, 3) Be ae ae OF On comparing the terms of the latter with (2-83) and (2-86), it is seen that one has precisely V(V* F) = VF + V x (V x E). This is a vector identity, usually written Vx (Vx BE) = V(V-F) — VF (2-88a) Equation (2-88a) provides a useful equivalence for V x (V x F), especially if the field F is divergenceless (V - F = 0), Then Vx(¥xF)=—-WF ifV-F=0 (2-88b)92 VECTOR DIFFERENTIAL RELATIONS AND MAXWELL'S DIFFERENTIAL RELATIONS ‘BLE 2-2. SuMmany of vector identities Algebreie Differential 1) F-G=G-F 1) V+) = Vit Ve 2) FxG=-GxF ay 1(F+G)=V-F4V-G F(G+H)=F-G+F-H (3) Rx (F+G)=VxFtVxG 4) Fx (G+H)=FxG+Fxi (16) VU) = f¥et VF 5) Fx (Gx HB) (HF) — H(F-G) (15) Ve (AR) = FWP (V- BE) F (Gx H) =G- (Hx F) =H: (Fx G) (16) V+ Fx G)=G- (Vx) -F-(V x G) anv WY) x FALV xB) Integral is) V+ af 7) PF-ds= | V-Fav (19) ¥ 6 if fi {20 Vx 8) 8) GF de~ [Wx Fas OL VeVaP)= VV-F) VP ) Vx Uf Vg) = Wx Vi 9) fpstve) ds =f [Eve + (WA) (Vall do ‘ . 0) PU Ve~aWsleds= [Ve 2 V9) de Although the proof of (2-884) was carried out in the rectangular system, such differ- ential results are independent of the coordinate system, meaning that (2-88a) and (2-B8b) are true for any system. Tt is worthwile to observe that one can more easily expand VF by use of the vecior identity (2-884) than by definition 2-82), Thus WF = V(V-F)—V x (V x F} (2-89) is useful for expanding V°F in a coordinate system other than the cartesian. Several vector identities involving the differential operators grad, div, and curl are listed in Table 2-2 along with vector algebraic and integral identities. Proofs of the algebraic and the differential identities are achieved in the manner used to prove (2-88a}, that is, expanding both sides in rectangular coordinates leads to an identity. ‘The integral identities (7) aud (8) are recognized as those of divergence and Stokes's theorem, respectively, Extensions of the divergence theorem lead to Green’s integral identities (9) and(i0), proved in the next section, 2-8 GREEN’S INTEGRAL THEOREMS: UNIQUENESS One can specialize the divergence theorem {2-34} to a particular class of veotor fune- tions and obtain the integral identities known as Green's theorems. Suppose F to be a scalar field f multiplied by a conservative vector field Vg; let F = f Vg. Then (2-34) takes on the special form hve) tas five Uf Vg) do (2-90)29 WAVE EQUATIONS FOR ELECTRIC AND MAGNETIC FIELDS IN FREE SPACE 93 assuming the functions well-behaved in and on the volume V. The integrand in the volume integral may be expanded by use of (15) in Table 2-2, whence (2-90) becomes Greon’s first integral identity Qi Ve) da =f LFV OY (Re) do 91) If one chooses to define a vector function G = gV/ instead, the same procedure leads to a result like (2-91) except for the interchange of the roles of the scalar functions f and g Gey -de= f ev4r4 val Ode Subtracting the latter from (2-91) obtains Green’s second integral identity Pr Ve Wf eds = [UV ev4pvde (2-92) also known as Green’s symmetric theorem. Green’s theorems (2-91) and (2-92) are important in applications to theorems of boundary-value problems of field theory, as well as to special theorems concerning integral properties of scalar and vector functions. One such, theorem concerns those dif ferential properties of a vector field F that must be specified in a region to make F uni- que. This theorem, not proved here,® shows that the specification of both the divergence and the curl of a vector function F in a region V, plus a particular boundary condition on the surface 5 that bounds V, are sufficient to make F unique. Maxwell’s equations (2-39), (2-41), (2-62), and (2-63) specify the divergence and the curl of both the E and the B fields in a region (in terms of charge and current densities as well as the B or E field}, so that these relationships, together with appropriately specified boundary conditions, can similarly be expected to provide unique field solutions. Finding solu- tions of Maxwell’s differential equations is facilitated for some problems by first mani- pulating them simultaneously to obtain differential equations in terms of only B or E, as is discussed next. *2-9 WAVE EQUATIONS FOR ELECTRIC AND MAGNETIC FIELDS IN FREE SPACE Electromagnetic field solutions B and E in free space must, by the uniqueness discus- sion of the previous section, satisfy the Maxwell divergence and curl relations (2-39), (2-41), (2-62), and (2-63). In a time-varying electromagnetic field problem, one is generally interested in obtaining E and B field solutions of the four Maxwell relations, a process that can often be facilitated by combining Maxwell’s equations such that one of the fields (B or E) is eliminated, yielding a partial differential equation known as the wave equation. This is accomplished as follows. *For a proof, sce S, Ramo, J. R. Whinnery, and T. Van Duzer. Fisldi and Waves in Communication Electronics, 2nd ed. New York: Wiley, 1985, p. 130. *For the purposes of the next section, Section 2-9 may be omitted if desired.94 VECTOR DIFFERENTIAL RELATIONS AND MAXWELI’S DIFFERENTIAL RELATIONS The Maxwell differential equations for free space are repeated here for conyen- ience. [2-39] ‘ [2-41] 2B E=—> 2-62 vx ai [2-62] a(€oE) a [2-63] To eliminate B, taking the curl of both sides of (2-62) obtains é Vx (Vx E)=—3 (Vx B) (2-93} Substituting (2-63) into the right side of (2-03) yields, after transposing terms contain- ing E to the lefi side Ee ay Vx (Vx B) + toto Gy = — Hoy (2-94) a vector partial differential equation known as the inhomogeneous vector wave equation for free space. ‘A wave equation similar to (2-94) can be obtained in terms of B. Thus, taking the curl of (2-63) and substituting (2-62) into the result yields the inhomogeneous vector wave equation eB Vx (V xB) + tigto ax = MoV XJ (2-98) From (2-41), B is always divergenceless, and with V+ E = p,/c, (2-94) and (2-95) are written, CE a. vp, WE = lof) sy = Mo 2 2-96 Hote az Moa e Tahomogencous vector (2-96) wave equations for 2B charge-free region WB — Ho€o Ga ~ —HoV x J * ® (2-97) A further simplification is possible if the region is empty space; that is, it is both charge free and current free (p, = J = 0). Then the simpler homogeneous vector wave29 WAVE EQUATIONS FOR ELECTRIC AND MAGNETIC FIELDS IN FREE SPACE 95. equations bold E V°E = toto Gr = 0 / (2-98) ———_——— Homogeneous vector wave equations for empty space 2 , oB VB — woes Sa = 0 (2-99) If in a problem the yeclangalar coordinate system is appropriate to the E and B fields governed by (2-98) and (2-99), making use of (2-83) provides the following scalar wave equations in terms of field components OE, VE, = feo GE = 0 (2-100a) ve, VE, ~ toto za" = 0 (2-100b} CE, VE, ~ Lo€o ae 7 0 (2-1000} with an analogous trio of equations in B,, By, and B, yielded by (2-98). The complex time-harmonic forms of the wave equations may be obtained by re- placing B and E with their complex exponential forms, (2-67). If this is done for (2-98) and (2-99), one obtains after cancelling e! VB + @7Uo€ok = (2-101) ————---- Homogeneous vector wave SEE equations in complex time-harmonic WB + cw Upeob form, for empty space (2-102) Since B =a,é, + a,£, + a,2, and B= a,B, + a,B, + a,B,, (2-101) and (2-102) expand to obtain the following homogeneous, scalar wave equations in complex time- harmonic form VE, + wUceok, = 0 (2-103) WE, + wo Ugeh, = 0 (2-104) V7B, + cw ptpepli, = 0 (2-105)96 VECTOR DIFFERENTIAL RELATIONS AND MAXWELL'S DIFFERENTIAL RELATIONS with a similar triplet of equations in B,, B,, and B, yielded by (2-102}. The simplest solutions of these scalar wave equations are uniform plane waves, involving as few as two field components. They are considered in the next section. +2-40 UNIFORM PLANE WAVES IN EMPTY SPACE | The simplest wave solutions of Maxwell’s equations are uniform plane waves, char- acterized by uniform fields over infinite plane surfaces at fixed instants. Simplifying fea tures are that the solutions are amenable to the rectangular coordinate system, and the number of field components reduces to as few as two. These simplifications provide a background for the more complex wave structures discussed in later chapters. Uniform plane waves have the property that, at any fixed instant, the E and B fields are uniform over plane-surfaces, These planes are arbitrarily chosen; for present purposes, assume that they are defined by the surfaces z = constant. This is equivalent to stating that space variations of E and B are zero over z= constant planes; thus assume 1. The fields have neither x nor_y dependence; that is, 2/4x = 0/0y = 0 for all field components. It will be shown that waves propagating in the z direction result from this restriction. If the waves propagate in empty space, one requires an additional assumption. 2. Charge and current densities are everywhere zero in the region; that is, Po= J=0. The complex time-harmonic forms of the Maxwell differential equations deter- mining the wave solutions are (2-70) through (2-73). With assumption (2) they become V> (eo) =0 (2-106) vV-B=0 (2-107) (2-108) (2-109) Combining these equations has been shown to produce the wave equations (2-101) and (2-102) VE + @2pgeo8 = 0 [2-101] WB + @p9¢08 = 0 [2-102] *This section on plane waves, plus Section 3-6 in Chapter 3, may optionally be omitted at this time, if desired, and taken up immediately before beginning Chapter 6. Plane wave concepts are included here because of their universal relevance to all dynamic field phenomena, and because they are essential to a more complete understanding of conduction and polarization effects in materials under other than purely static conditions.210 UNIFORM PLANE WAVES IN EMPTY space = 97 One should bear in mind that no new information is contained in the latter that is not already expressed by the preceding Maxwell’s equations. Before attempting to extract solutions from the wave equations, one may note that the curl relations, (2-108) and (2-109), furnish some interesting properties of the solutions, restricted by assumptions (1) and (2). Assuming that all six field components are present, (2-108) becomes, with G/ax = 2/dy = 0 of assumption (1), . la, a, a,| VxE=|0 0 E, a - “ i « E> wioaB. + 0B, + a8) £, bp expanding into the triplet of differential equations (2-110a) (2-110b) o=B, (2-110c) Similarly, (2-109) provides a 3, Ho. gp Get loeoks (2-11 la) (2-111b) o=k, (2-11 1c) From these differential expressions, the following properties apply to the solutions about to be found 1. No £ component of either B or B is obtained, thus making the ficld directions entirely transverse to the z axis. 2. Two independent pairs of fields, (Ex, B,) and (E,. B,), are yielded under the as- sumptions. This is seen to be the case on setting £, = 0 in (2-110b), for example, forcing B, to vanish while yet leaving the field pair (£,, B,) intact, the latter being governed only by (2-1 10a) and (2-111b), When field pairs are independent of each other, they are said to be uncoupled.98 Vecror DIFFEREWTIAL RELATIONS AND MAXWELL'S DIFFERENTIAL RELATIONS Suppose one desires wave solutions involving only the field pair (E,, 2,). Then put £, = B, = 0, reducing the pestinent differential equations to just (2-1 10b) and (2-11 fa} chs —joB, [2-110b] ‘ abs - — joe ob, [2-111a] ze The field solutions arc obtained on combining (2-110b) and (2-11 1a) to eliminate E, or B,, yielding a scalar wave equation from which solutions can be found. Alter natively, one can make use of cither vector wave equation (2-102) or (2-103), sub- Jecting it to the same assumptions. (Only £, and B, are present and a/éx — o/éy = 0.) Either approach obtains the following wave equation in terms of E,: Ps 62 + wugegl, = 0 (2-112) ‘This is a partial differential equation in one variable (z); thus it can be written as the ordinary differential equation PF, ee + 0 p1€ 8. 0 (2-113) Its solution is the familiar® superposition of two exponential solutions Ex(o} = Eye P* 4 Celto (2-114) wherein @, and @, are arbitrary (complex) constants and the coefficient By, called the phase constant, ig given by By = ©/uo€p. It is to be shown that the exponential solutions Ce" J" and C0! are representations of constant amplitude waves traveling in the posi- tive z and negative z directions, respectively. The complex coefficients C, and C, must have the units of volts per meter, denoting arbitrary complex amplitudes of the positive z and negative z traveling waves. Employing amplitude symbols &t and Ey instead of @, and €,, puts (2-114) into the form Bz) = BgeW iP" 4 Be ello Vim = Ba) + B28 {2-115} ‘The complex amplitudes £;) and EZ may be represented by points in the complex plane using the Argand diagram of Figure 2-10, so from their polar representations ExSmye** and By = Bel (2-116) with $* and @~ denoting arbitrary phase angles. “Lt is assumed that the reader is familiar with the details of this solution, found in any text on ordinary differcntiat equations210 UNIFORM PLANE WAVES IN EMPTY space 99 (Complex plane) FIGURE 2-10. Complex amplitudes rep- resented in the complex plane. Once a solution of one wave equation has been obtained, the remaining ficld can be found by use of Maxwell's equations. Thus, the solution (2-115) for &,{z) inserted into the Maxwell relation (2-110b) yields ~ep- Ea Bo + Bi elBo=] = Viitgeobth cP — Sigegfzel#? Wh/m? (2-117) in which Bo once more denotes the space phase factor iggy vad/m (2-118) ‘The real-time, smusoidal steady state expression for the electric field component is found from (2-74). Taking the real part of (2-115) after multiplying bye” obtains Eyles) = Re [E(ahe*!] = Re [(Bjelbt ¢Hlos 4. Br pt pitas) py = Ex, cos (wt — Boz + O°) + By cos (t+ Boz + 7) — (2-119) Note that Ht and Bj, denote the traveling wave real amplitudes, whereas 6* and od” are arbitrary phases relative to the instant ¢= 0 and the location z = 0 in space. ‘The real-time form of B, of (2-117) is similarly found to be 251) = [VMo€oZg | cos (ot — Boe + 6*) — [vHo€ofin | cos (wt + Boz +o } (2-120) The traveling wave nature of uniform plane waves can be grasped from a graphic interpretation of (2-119} and (2-120). Consider only the first terms of each: the positive z traveling wave. The following symbols are chosen to denote them. CES (2, 1) = Ex cos (mt — Bye + 6*) Vin (2-121a) By (%, = Vite okin cos (wt — Boz + b*) Wh/m? (2-121b) ‘Their positive z traveling nature may be observed if (2-121) is plotted as a family of cosine waves versus 2, at successive instants of time &, (When observing the effects100 vector DIFFERENTIAL RELATIONS AND MAXWELL’S DIFFERENTIAL RELATIONS Ef (zt) = Em cos wt — Boz + o*) ‘atta 0 Wave motion, with increasing ¢ (a) (x) z (or Boz) ———> Motion, (b) FIGURE 2-11, Electric field sketches of a positive z traveling uniform plane wave. {a) Vector plot along ¢ at successive instants. (5) Flax plot of the electric field at ¢ = 0. of time or space variations of a field, it is usually best to hold space or time fixed, while the other is allowed to vary.) At t=0, (2-12la) becomes EZ (z, 0) = Ey, cos (~ Boz + 6*) = Ex cos (Boz — 6*), shown plotted against the z variable as the solid line in Figure 2-11(@). With the period T defined by (2-122) at one-eighth period later, for example, (2-121) becomes Ey (z, T/8) = Eq, cos (Boz — 20/8 — *). The cosine function is thus shifted in the positive z direction by the time lapse of the eighth period as showa, yielding a positive z motion of the wave with increasing time. The vector field plot of Figure 2-11 (a} shows only a, (2, #) along a typical z-axis in the region. To display the field throughout a cross section in any x-z plane, the flux plot of Figure 2-11(b) is more suitable. The_motion of the wave with increasing ¢ is related to the phase factor Bo = @y/Ho€o appearing in the wave expressions, with Boz having the units of radians (dimensionless), implying that Bq is given in radians per meter. The z distance that the wave must travel such that 2x rad of phase shift (one complete cycle) occurs is called the wavelength, designated by the symbol 4 and defined by Bod = 2m rad (2-123)2.10 UNIFORM PLANE WAVES IN EMPTY space 101 Thus, the wavelength in free space is related to the phase factor By by Qa Qn Bo” oviteca An observer moving with the wave such that he experiences no phase change will move at the phase velocity of the wave, denoted by v,. The equiphase surfaces of the positive ¢ traveling wave are defined by setting the argument of (2-121) equal to aconstant; that is, a ~ Byz + @* = constant, whereupon differentiating it to evaluate dzjdt yields the phase. velocity (2-124) dz @ op Ga piles (2-125a) Because By = W/jigeg, and with to = 4x x 1077, €g ¥ 107 9/36z, the phase velocity of a uniform plane wave in empty space is 1 VEoeo 3 x 10° m/sec (2-125b) the speed of light.” . . _ A comparison of the complex expressions, (2-115) and (2-117), for £,(z) and },(@) shows that their separate traveling wave terms are paired into ratios producing the same constant. Thus, write (2-115) and (2-117) in the forms Eka = Ee itor + By, elbor = Ex (2) + Bote) (2-126) and (2-127) in which £3 (2), £7 (z) and BY (x), By (z) symbolically denote the positive z and negative z traveling wave terms directly above’ them. Then the following complex ratios hold at any point in the region BO BO 1 gs ; Bw & Vimeo =X m/sec (2-128) to provide a means for finding one of the fields whenever the other is known, A more common variation of this technique is achieved by modifying the B field in empty “Experiments have shown that the speed of light, is mare nearly 2.99792 x 10® mysec. This value, together with the assumed permeability for free space pug it x 10-7 H/m, inserted into (2-125b), is seen to lead a value for €g that departs slightly from the approximate value 10° 9/36x given.102 vector DIFFERENTIAL RELATIONS AND MAXWELL'S DIFFERENTIAL RELATIONS Ed (2.1) = Eh, cos (at — Boz + 3*) Equiphase surface By tz,t)= 72 cos (ut - Boz + 6") (atr=0) Eg flux FIGURE 2-12. Vector plot of the fields of a uniform plane wave along the z axis. Note the typical plane equiphase surface, depicting faxes of BE and 2 space through a division by flo, defining a magnetic intensity field denoted by the symbol H for empty space as follows. B gg 7H Mm For empty space (2-129) 0 Thus, denoting BF (z)/Ho by Af (<), and By (z)/tt by Ay (z) the following ratios of the traveling wave terms are valid for plane waves in empty space EXX2) = =o = 12002 (2-130a° Bie) "0 (21308) (2-130b) The real ratio, y/{ig/€y (having the units volts per meter per ampere per ineter, or ohms), is called the intrinsic suave impedance for empty space, and is denoted by the symbol 9. The advantage of (2-130) over (2-128) is that the ratio 7 is a usefully smaller number. The real impedance ratio of (2-130) shows that the electric and magnetic fields of uniform plane waves in eapty space are in phase with one another, a condition evi- dent on comparing the positive ¢ or the negative z traveling solutions of (2-126) and (2-127). Each contains the same phase argument in the exponential factors, ample evidence of their in-phase condition. Figure 2-12 depicts the real-time electric and magnetic fields of (2-121) plotted against z in space at t= 0. EXAMPLE 2-44. Suppose a uniform plane wave in empty space has the electric field Biz) = 01000? Vim a)211 WAVE POLARIZATION 103 its frequency being 20 MHz. (2) What is its direction of travel? Its amplitude? Its vector direction in space? (6) Find the associated B field and the equivalent Hi field. (<) Express §, B, and H in real-time form. (d) Find the phase factor Bo, the phase velocity, and the wavelength of this electromagnetic wave. (a) A comparison of (1} with (2-115) or (2-126) reveals a positive z traveling wave, whence the symbolism: Ez} = a,B7(z). The real amplitude is BE = 1000 V/m, -with the vector field x directed in space. (0) Using cither (2-127) or the ratio (2-128) (2) = * 000 BQ) = = oT HOF = 8.38 x 107 Se-H89® Wh/m? 7 ‘The use of (2-130a) obtains the magnetic intensity E() _ 1000 199! 9. 659~ Jo a aoe’ 2.656280 A/mm Aj (2) = () The real-time fields are obtained from (2-74) by taking the real part after multi- plication by e* EX (z, 8) = Re [1000e~#*el@"] = 1000 cos (@ — Boz) Vim. BS {z, t) = 3.33 x 107° cos (cot — gz) W/m? (or T) HBS (z, 0) = 2.65 cos (of — Boz) Afm {d) Using (2-118), (2-125), (2-124), and (2-122) yields _ _@._ 2n(20 x 109) Ba ON 69 = TH = 3 x 10® m/sec 2a 3x 108 By fF 20x 108 0.42 rad/m a, A 15m 2-44 WAVE POLARIZATION The vector orientation, or polarization, of an electromagnetic wave in space is usually described with reference to its electric field direction, Thus in Figure 2-12, the z traveling uniform plane wave shown with the field components E,, If, is said to be polarized in the x direction (or simply x-polarized). Similarly, the plane wave with the components Ey, H, described in Problem 2-51 is polarized in the y direction. Both these waves are linearly polarized, because the electric field vector in any fixed z plane describes a straight-line path as time passes. Because Maxwell’s equations are linear equations, a vector superposition, or summing, of the two linearly polarized uniform plane waves just introduced will also provide a valid field solution, The resultant vector sum will not necessarily be linearly polarized, however, depending on the phasc condition betwecn the x and the y- polarized electric field components. For example, with E,(z, 0 = Eqs 608 (en! — Boz) and E,iz, f) = Egy cos (cot — Boz) propagating in phase and at the same frequency along the z-axis, their sum, E=a,£, +a,£,, would appear as depicted in Figure104 = vecTor DIFFERENTIAL RELATIONS AND MAXWELL’S DIFFERENTIAL RELATIONS Locus of E in 22:0 plane wey, 00 @ (a) Locus of E in2=0 plane @ 20.8) r f Bife,03= By, 00Skot Boe) ret Ee, =B yy Comet Bp + 90°) \ #08 Locus of total tb) FIGURE 2-13. Uniform plane waves shown in space at ¢= 0 and added to produce different wave polarizations. (a) Cophasal , and 2, their sum yielding a linearly polarized result. (8) Elliptical polarization of E produced with 90° phasing, (Related H field componcats are omitted for clarity.) 2-13(a), producing a linearly polarized field E, tilted in any fixed z plane by the angle = arc tan (Eqy/Eme) from the x-axis, as shown. The equation of the straight line may be found ‘by conveniently inserting z= into the E, and E, expressions and forming their ratio to eliminate ct, yielding Eny gods (2-131) ‘This is evidently the equation of the straight line (form: y = mx) as shown in the inset diagram of Figure 2-13(a), regarding By and E, as the variables in lieu of x and y.REFERENCES 105 On the other hand, if the two omponent fields were 90° out of phase, such that Ey = Ey, 608 (Cot — yz) and E, = Emy 608 (ext ~ Boz + 90°) as in Figure 2-13(8), the sum E—a,P, + a,B, would produce the spiraling locus of the E vector about the zaxis as noted. In the fixed z=0 plane, the. component fields are written FE, = Eqoe C08 Of and Ky = yy C08 (Ct + 90°) = — Ey Si. Of = = Eyy afl — cos? ot. "Ine serting the £, expression into E, to eliminate ct yields the locus of E in the z = 0 plane. (2-182) the equation of an ellipse with principal axes of half lengths E,, and Ey), as seen in Figure 2-13(6). Thus, the tip of the total E vector describes an elliptical locus in any fixed z plane as the wave moves by, indicating the elliptical polarization of the wave. It is also evident that a circular polarization of the E vector would occur if Ez = Emp in. (2-132), You may show that if the 90° phase condition between E, and E, were replaced by the general angle 8, making E, = Fy, cos (ot + 8) in the <= 0 plane, then the polarization locus would acquire the form of ER EE cos 4 Enz Ey Erm E ny E,Ey — sin? 6=0 (2-133) an ellipse with its major axis tilted, depending on the choice of 0. Wave polarization is of practical importance in radio communication transmit- receive links because the power extracted by a receiving antenna from the arriving wave is usually dependent on the orientation of the antenna relative to the polarization of that wave. The common half-wave, thin wire dipole antenna, for example, picks up the maximum power from a linearly polarized oncoming wave when the electric field of the arriving wave is aligned with the antenna wire, while accepting zero power from the wave if the electric field and the wire are at right angles. If the arriving wave were circularly or elliptically polarized, a component of the arriving E-field vector is made available to the receiving dipole regardless of its tilt in the plane of E, so that the orientation of the receiving antenaa, in any fixed c-plane, would have little or no effect on the amount of signal received. This could be of considerable importance in satellite communications, in which the receiving antenna on the satellite is tumbling in space and therefore changing its attitude relative to the oncoming wave. Antennas capable of transmitting circularly polarized waves, such as helical antennas or phased crossed dipoles, are readily constructed to accommodate this need. FERENCES Asranam, M,, and R. Beoxsr, The Classical Theory of Electricity and Magnetism. Glasgow: Blackie, 1943. Fano, R. M., L. T. Giu, and R.P. Apuer. Electromagnetic Fields, Energy and Forces. New York: Wiley, 1960. Kraus, J.D. Electromagnetics, 2nd ed. New York: McGraw-Hill, 1984. Parciaes, HB. Vector Anaiysis. New York: Wiley, 1944. Ramo, S.,,J. R- Wainnery, and T. Van Duzer, 2nd ed. Fields and Waves in Communication . New York: Wiley, 1984.106 vecror DIFFERENTIAL RELATIONS AND MAXWELL’S DIFFERENTIAL RELATIONS, PROBLEMS SECTION 2-2 2-1. From the substitution of the appropriate coordinate variables and metric coefficients into the gradient expression (2-10), show that (2-Ia,b,c) follow in the three common coordinate systems, Also convert the magnitude expression (2-13) to correct forms in. those systems. 22. Express as a vector function the gradient {maximum directional derivative) of the fol- lowing scalar fields (a) f(x) = 20x; (b) ax, y, 2) = 20x? + 30xy? + Haye (c) F(r) = 100/75 (d) lp, &, 2) = 5p sin — 6p?z cos ; (c} Ar, 0, ) = 0/7 + 10r cos O + 207? sir O sin 2. [Answers (b) a,(40x + 80)? + 40yz) + a,(60xy + 40xz) + a,40zy (d) a, (5 sin @ — 12pz cos $) + a,(5 cos @ + Bpz sin @) — a,6p? cos @] 23. Prove, by expression in rectangular coordinates, that ¥(f + 2) = Vf + Vg (vector iden- tity (11) in Table 2-2), 24, With f and g given to be scalar, differentiable fields, by expansion in rectangular coordi- nates prove the identity {14) in Table 2-2, that V(/g) = f Vg + Wf. 2-5. In Problem 1-6 is depicted the “‘distance-vector,” R, defined as the difference #, — r, of the position vectors to the endpoints of R. Relabel the point P2 now as P(x, y, <) with arbitrary coordinates (taken to be differentiation variables), so that now R = x - x1. (a) Write the expres- sion for R as well as its magnitude 2 in rectangular coordinates. (b) Show that WR is also the unit vector in the direction of R. SECTION 2-4 26, Carry out a direct proof resembling that leading to the expression (2-28) for div F, but carried out in the rectangular coordinate system. Begin with (2-22), expressed in rectangular coordinates with reference to a diagram like Figure 2-4 but adapted to the rectangular system. 27. By the substitution of the appropriate coordinate variables and metric coefficients into (2-28), show that the expressions (2-29a,b,c) follow, in the three common coordinate systems. 2-8. Determine for each of the following vector fields whether or not it has fiux sources; that is, find its divergence. (a) A= Sa, + 4a, (constant vector field in a region) (b) F(x,y, 2) = Sxza, + daya, + (5x? + 3)a, (c) Glx,y, 2) = Syay + dea, + (537 +9) @, (4) H(x,», 2) = 6sa, + 6a, + 6za, = 6a,r (determine it in both rectangular and spherical coordinates} (e) Jp. $2) = a,Spz sin @ + ayl0pc cos {f) K¢r, 0, @} = a,100/r? + 2520/7 + agl0r cos @ [Answer: div F = 3z + 4s, fields A, G, and J are sourceless] 29. Prove, by expansion in rectangular coordinates, that V+ (F+G) = V-F+¥-G, the identity (12) in Table 2-2. 210. By expansion in the rectangular coordinate system, prove the identity (15) in Table 2.2, V+ PR) = FW + fi 8). 241, Show that the following fields are divergenceless (source-fee). (a) The p-directed, inverse-p dependent field F = a,/p, for p > 0; and (b) the redirected, inverse-r* dependent field, G = a,/r?, for r > 0. (By comparison with results found in Example 1-11, with what kinds of static-charge sources are these field-types identified?) 212, (a) Given the class of electric fields E(p) = a, K/p" with Ka constant and na parameter, find div E. What choice of n yields a divergenceless (charge-ftee) field everywhere (excluding p =0)? Comment on this conclusion relative to (1-61}, applicable to the uniform line charge. (b) Given the class of electric fields E = a,K/r", find div E for > 0. Which choice of the parameter n provides a divergenceless field? Comment on this conclusion with respect to (1-57b), the electric field of the point chargeprosLems 107 SECTION 2-44 213, Assuming the same six-sided closed surface § to bound the box-shaped interior volume as in Example 2-4, assume the field G(x, 9, 2) = 910x972? exists in the region. Iilustrate the validity of the divergence theorem (2-34) by evaluating its volume and surface integrals in and on the given parallelepiped. [Answer: 10,080] 2-14, Assuming the same right circular cylindrical tegion of radius p = a and length ¢ as for Example 2-5, illustrate the correctness of the divergence theorem for this region, given the clectric field E = a,p9p7/4€o2”, that corresponds to the nonuniform charge density of Problem 1-48. {Note for this case that no singularity exists within the given V or on S, thereby obviating any need for the exclusion surface S; used in Example 2-5.] [Answer: 2Lp07/2€9] 215. The first octant of a sphere centered at the origin is bounded by the four coordinate surfaces: r= a, 6 = 0, ¢ = 7/2, and on the bottom by the plane @ = 7/2. Sketch it, Given that the ficid Fy, 8, 6) = a,10 — a430r sin cos @ exists in this région, illustrate the truth of the divergence theorem (2-34) by evaluating the volume and surface integrals within and on the defined region for the given field. [Answer: 1027(w + 7/2)] SECTION 2-48 2.46, In Problem 1-28, the electric field within the uniformly charged spherical shell (a (8-324) No dielectric material is strictly linear in its electric polarization behavior, though many are very nearly so over wide ranges of applied E fields. If E is made strong enough, a material may experience polarization displacements that result in permanent dislocations of the molecular structure, or a dielectric breakdown, for which case (3-25) does not hold. In a nonlinear material the magnitude of D is not proportional to the applied E field, {though the E and the P vectors may have the same directions). Then (8-25) is written more generally P= 7(E}eoE (3-33) in which the dependence of x, on E is noted.3-2 ELECTRIG POLARIZATION AND DIVD FOR MATERIALS = 12H A. Dielectric Polarization Current Density If the electric field giving rise to diclectric polarization effects is time-varying, the resulting polarization field is also time-varying. Then the displacements of the positive charge constituents in one direction, together with the negative charges moving oppositely, give rise to charge displacements through cross sections of the material identifiable as currents through those cross sections. Applying a time-derivative operator to the p; terms of (3-16) thus yields a current density interpretation as follows XS ad; . 2 me Eo Ay Ae ot) 7 (3-34) The resulting time derivative of the polarization field, 6P/ét, having the units of volume current density, is given the symbol J, as follows P Jp = Afm? (3-35) et and is called the electric polarization current density. The field J,, along with the polariza- tion charge density field p, described by (3- 20, acts as an additional source of electric and magnetic fields. In particular, the special role played by J, in relation to magnetic fields in a material region is discussed later in Section 3-4. B. Integral Form of Gauss’s Law of Materials The dielectric polarization effects attributed to material regions have been seen to lead to the divergence expressions (3-21) and (3-24), relating the field quantities P and D to the polarization charge and free charge sources. The divergence theorem can be used to transform these differential equations into corresponding integral forms. ‘The most important of these is (3-24) for the D field; that is, ¥-D = p,. Multiplying both sides of (3-24) by dv and integrating throughout an arbitrary volume region 7 yields [¥-Ddo= f ppdo (3-36) By the divergence theorem (2-34), the left side can be replaced by a closed-surface integral to yield § D-ds =f p,doG (3-37) in which S bounds V. Equation (3-37) is the inlegral form of Maxwell’s equation (3-24) for a material region, sometimes called Gauss’s law for material regions. Ut states that the net outward flux of D over any closed surface is a measure of the total free charge contained by the volume V bounded by S, at any instant of time. As expected, it be- comes the free-space Gauss’s law (1-53) if 7. = 0, reducing D to €gE.122 MaxwELL’s EQUATIONS AND BOUNDARY CONDITIONS Another divergence relation, V-P = —p, of (3-21), has the equivalent integral form de: ds=—J ppde€ (8-38) obtained by the method analogous to that used in Converting (3-24) to Gauss’s integral (3-37). Equation (3-38) states that the net outward flux of P emanating from the sur- face of V is a measure of the net polarization charge summed throughout V. C. Spatial Boundary Conditions for Normal D and P In many electromagnetic field problems of physical interest, it becomes necessary to discuss how the fields behave as one traverses the boundary surfaces, or interfaces, separating the various material regions that comprise the system. In such problems, a matching or fitting of the field solutions is required so that the boundary conditions at the interfaces may be satisfied. The proper boundary conditions for the fields are determined, as will be shown, from the integral forms of Maxwell’s equations for material regions. ‘The Maxwell integral relation (3-37), 45 Dds = J, p, dv, can be used, through an appropriately constructed closed surface, for comparing the normal components of D that appear just to either side of an interface separating two materials of different permittivities. Denoting the materials as region | and region 2 with permittivities €, and €, define a pillbox-shaped closed surface of small height 54 and end arcas As so that both regions to either side of the interface are penetrated as in Figure 3-4. Calling the fields D, and D, at points just inside regions 1 and 2, respectively, the application of the left-hand integral of (3-37) to the closed pillbox yields the net outward flux from the top and bottom surfaces As. At the same time, the right side is the charge enclosed by the pillbox; this is p, Asdh, so (3-37) becomes Dyy As ~ Dyp As = p, Ash (3-39) The right side of (3-39) vanishes as dh + 0, assuming p, denotes a volume free charge density in the region. If however, a surface charge density denoted by p, and defined by the limit p, = lim p, 5h (3-40) ano Region 2: @) De (a) (2) FIGURE 3-4. Gaussian pillbox surface constructed for deriving the boundary condition on the normal component of D. (a) Pillbox-shaped closed surface showing total fields at points adjacent to interface. (6) Edge view of (a), showing fields resolved into components.4.2 ELECTRIC POLARIZATION AND DIV D FOR MATERIALS 123 js present on the interface, (3-39) reduces to the general boundary condition Dyy ~ Dyz = 2, Cfm™ (3-41) Equation (3-41) means that the normal component of D is discontinuous to the extent of the _free surface charge density present on the interface. Since Dyy = m*D, and Dyz = a+ Do, with ‘a denoting a normal unit vector directed from region 2 toward region | as in Figure 3.4(b), (3-41) is written optionally in vector notation as follows. 4 — D2) = p, Cfm (3-42) ‘The boundary condition (3-41) is true m general, but for some physical problems a free surface charge density p, may be absent. Two special cases of (3-41) of physical interest are mentioned in the following, while a more general result is left for discussion in Section 3-L1. Cast A. Both regions perfect dielectrics. A perfect dielectric, for which the con- ductivity o is zero, cannot furnish free charges, so that if no excess charge is supplied to the interface by an external agent (rubbing it with cat’s fur, for example), then p, = 0 on the interface. Then (3-41} reduces to Days = Dy Gime’ (3-43) The normal component of Dis continuous at an interface separating two perfect dielectrics, as illustrated in Figure 3-5{a). Region 1: (a1 = Ore) Region 1: («)) Interface Region 2: (03 = 0; 2) Region 2:(¢9—~00) @ O) FIGURE 3-5. Two cases of the boundary condition for normal components of D. (a) Continuous D, at an interface separating perfect dielectrics. (#} Equality of normal D, to a surface charge density on a perfect conductor.124 MAXWELI'S EQUATIONS AND BOUNDARY CONDITIONS CaskB. One region is a perfect dielectric; the other is a perfect conductor. Electric currents are limited to finite densities in the physical world. Thus from (3-7), J = GE, the assumption of a perfect conductor in region 2 of Figure 3-4 (6) > 00), implies that E, in that region must be zero if the current densities are to have, at most, finite values. Moreover, with electromagnetic fields satisfying (2-108), V x £ = —joB, one can see that if E, is zero in region 2, then 'B, must be zero there also. Thus for time-varying fields, o+0 implies E=B=0 (3-44) in a perfect conductor. The boundary condition (3-41) or (3-42) then must reduce to Day = Ps, OF in vector form n-D =p, C/m? (3-45) The surface charge density residing on a perfect conductor equals the normal component of D there, as illustrated in Figure 3-5(4). . In a static field problem involving only jixed electric charges and no static currents, the boundary condition (3-45) holds true even though region 2 may be only finitely conducting, for the assumption of no static currents in the finitely conducting region 2 implies from (3-7) that E, = 0 there, making D, = 0 as well. Thus (3-42) reduces to (3-45). ‘A boundary condition similar to (3-41) can be derived comparing the normal components of the dielectric polarization vector P. Noting the similarity of Maxwell's integral law (3-37) and the polarization field integral (3-38) and using another pillbox construction, one can show that P,,; — P,2 = ~ Psp, or in vector form n+ (PB; —P2) = ~p., C/m? (3-46) in which p,, denotes the net surface bound charge density lying within the pillbox. "The net density includes the effect of both species of surface polarization charge (positive and negative) accumulated just to either side of the interface. A simpler picture is obtained if region 1 is free space, for which 741 = 0 (or €, = o). Then P, = 0, reducing to the special case B+ Py = py Clim? (8-47) The surface polarization charge density residing at a free-space-to-dielectric interface equals the normal component of the P field there. EXAMPLE 3-2. Two parallel conducting plates of great extent and d m apart are statically charged with +9 Con every area 4 of the lower and upper plates, respectively, as noted in (a). The conductors are separated by air except for a homogeneous dielectric slab of thickness cand permittivity ¢, spaced a distance from the lower plate. (a) Use Gauss’s law (3-37) to establish D in the three regions. Sketch the flux of D. (5) Find E and P in the three regions and show their flux plots. (c) Determine p, on the conductor surfaces, p, in the dielectric, and pp, at y= 0 and y=5 +6.3-2 ELECTRIC POLARIZATION AND DIV D FOR MATERIALS 125 Ly A, Negatively *; “~charged surface (-g/A) |e Dielectric slab L | Region 1 Positively | eaten i charged surface i afb te tL A dh ~ Y d t q oe % Voor (GIy Ay @ | NY caussan surfaces Area A- (rectangular boxes) (2) Moe A } een Psp We PL lip lll Lia HE Ps (e) @ @ EXAMPLE 3-2. («) Charged parallel conductor system. (b) Flux of D. (¢) Flux of €)E. (d) Flux of P, {a) E exists only between the conductors and by symmetry is independent of x and <. A Gaussian closed-surface § in the form of a rectangular box is placed as in Figure 1-15(d), to contain the free charge g. With static E inside the conductor zero, 2 D flux ofa constant density emanates from the top of 8, making the left side of Gauss's law (3-37) become 4 D- ds fi, (Dy) ayds=D, [La = Del Equating to the right side of (3-37), the free change ¢ = D,A, whence 4 D=a,D,=a,7 am a result correct for all three regions between the conductors because no free charge exists in or on the dielectric. The flux plot of D is shown in (6) (0) Eis obtained using (3-30c), so in the dielectric slab, D bey € for a typical dielectric, E in the air regions exceeds the value in the dielectric, as shown in (c). P in the dielectric is found by use of (3-23) p=p-gtwa(t-od)) a(@-) SG) © For ¢, > 1, P in the slab is positive y directed, as shown in (@). In air, P is zero. From (3-35), no polarization current density J, is established in the dicleetrie be- cause the fields are time-static. (c) The free charge densities on the conductors ate obtained from (3-45), yielding p.= +4/A. The polarization charge density p, from (3-21) is zero because P is a constant vector throughout the slab. The surface polarization charge density p,p is found by inserting (4) into (3-47), yielding Py = Psp ‘These surface densities are noted in (8) and (d) of the figure. 3-3 DIV B FOR MATERIALS: ITS INTEGRAL FORM AND A BOUNDARY CONDITION FOR NORMAL B In Section 3-2 the Maxwell relation for V+ D in a material was developed by adding the effect of the electric polarization charge density p, to the free-space Maxwell relation. The form of the expression for V- B in a material can be developed analo- govsly. No additive term is required in this case, however, because no free magnetic charges exist physically in any known material. ‘Thus B remains divergenceless in materials; that is, V+B=0Wb/m (3-48) Equation (3-48) is converted to its integral form using a technique analogous to that employed in obtaining (3-37). Multiplying both sides of (3-48) by do, integrating it throughout an arbitrary V, and applying the divergence theorem GB: ds =0Wb (3-49)$44 MAGNETIC POLARIZATION AND GURL H FOR MaTeRiALs 127 the integral form of (3-48). Equation (3-49) states that the net outgoing flux of B over any closed surface § is always zero, implying that B flux always forms closed lines. ‘A boundary condition concerned with the normal components of B and analo- gous to (3-42) can be found by applying (3-49) to a vanishing Gaussian pillbox like that of Figure 3-4. The resulting boundary condition is ‘ (3-50) that is, the normal component of the B field is continuous at an interface separating two adjacent regions. 3-4 MAGNETIC POLARIZATION AND CURL H FOR MATERIALS ‘The magnetic properties of a material are attributed to the tendency for the bound currents, civculating on an atomic scale within the substance, to align with an applied B field. Three types of bound currents are associated with atomic structure: those attributed to orbiting electrons, and those associated with clectron spin and with nuclear spin. Each of these phenomena, represented in Figure 3-6(a), is equivalent to the circulation of a current J about a small closed path bounding an area ds, the positive sense of which is related by the right-hand rule to the direction of J as in Figure 3-6(). The product Ids defines the magnetic moment sm contributed by those bound currents of the atomic or molecular configuration. It is shown that applying an external magnetic field B to the typical moment m = /ds yields a torque exerted ona, tending to align mm with the applied B field, One can in this manner explain the magnetic behavior of a material as though it were a collection, in empty space, ofmany magnetic moments m per unit volume. The tendency to align with the applied B field is shown to provide an equivalent magnetization current of density Jn, serving to modify the magnetic ficld in a certain way. A description of this process, beginning with a discussion of the torque produced by the B field on a current clement, follows. "A current loop of microscopic size has an external magnetic field behavior in- dependent of its shape in a plane, so a square loop is assumed in Tiew of the circular configuration of Figure 3-6(6). Itis shown in Figure 3-7(a) in the z = Oplane, immersed in the applied field B = a,B, + a,B, + a,B.. The Lorentz force acting on each of the four edges of the square current [oop is obtained from (1-52) dF, = dq x BN (3-51) ‘Orbital motion Electron spin vector ala vector Nuclear spit meres vector 1 feds Electron erbia, moter Nucleus. (a) @) FIGURE 3-6, The elements of bound currents that exist in atomic structure. (a Constituents of circulating currents associated with particles ofa simple atom. (3) Magnetic moment m of a current circulating about an arca ds.128 — MAxweEL1’s EQUATIONS AND BOUNDARY CONDITIONS dgy = Ide, ‘ FIGURE 3-7. Development of torque-expression for 2 current loop immersed in a B ficid. (a) Current loop immersed in arbitrary B field. (5) A moving charge element, dye, of the loop. (6) Development of torque @¥ produced on edge dé. if the charge dg moves with a velocity v along the edges dx and dy, One may cast (3-51) into the following forms, noting that dg = p, dv = p, df ds from Figure 3-7(0), and using (1-50a) Fy = pod dw x B= (Jaf ds] x B= Ldé xB (3-52) with the direction denoted by assigning a vector property to each edge length dé. The origin of the torque arm R is for convenience taken at the center of the loop. Along 4,, the differential torque dT, is given by Ry * dF, (Example 1-4), yielding aedy 2 “dT, =Ri x a,-|29| x [(-a,Jdx) x B] = —aJB, with the same result obtained for edge 3, while that acting on ¢, and ¢, becomes dT, + dT, = a,/B, dx dy. Thus the torque on the complete loop becomes i and with Zds denoting the magnetic moment a0 = (a,B, —a,B,)1 dedy = (a, x B)lds = Ha, ds) x B= Ids x B Ids Am? (3-53) one may abbreviate the result av =mxBN-m (3-54) It is clear from (3-54) that only the components of the applied B field in the plane of the current element act to produce a torque on it. [fan and B were parallel, dT would become zero; thus the torque @T is such that it fends to align the current element with the applied B field. ‘A very large number of current loops like those of the atomic model in Figure 3-2 comprise a magnetic material, susceptible to such magnetic alignment effects. In the absence of an applicd B field, they possess random orientations accompanied by thermal$4 MAGNETIC POLARIZATION AND GURL H FOR MATERTALS 129 x Sym = Mdo a @ 6) FIGURE 4-8. Current loop constituency of a maguetizable material, affected by an ap- plied B field. (2) Random magnetic moments, in the absence of B, (6) Partial alignment of magnetic moments, B applied. agitation effects, as depicted in Figure 3-8(a), if'one may avoid the subject of permanent magnetism occurring in some materials. bmpressing a B field develops a torque on each current loop, as specified by (3-54), such that the loops tend to align more or less in the direction of B as depicted in Figure 3-8(0). ‘The magnetization density M is defined in essentially the way the dielectiie polar- ization field P is defined by (3-16), that is, by summing the magnetic moments ma within a volume-element Av and expressing the sum on the per-unit-volume basis x am M= ae Ajm (3-55) This becomes a smooth functional result if the number W of current elements within Apis quite large, while Avis yet small enough to be considered suitable for manipulation in differential or integral expressions, Thus M furnishes a characterization of the circulating atomic currents within matter from a smoothed-out, macroscopic point of view. An important derivative function of the magnetization field Mis its curl, shown in the following to yield a volume density J, of uncanceled bound currents within a130) MAXWELL’S EQUATIONS AND BOUNDARY CONDITIONS magnetic material according to Jn =V x MAjmn? (3-56) r ‘A formal derivation of (3-56) proceeds with the aid of Figure 3-9. The examina- tion of an incremental volume-element of a material, depicted in Figure 3-9(a), reveals the presence of surface current contributions on Av as in (b) of that figure, assuming for the present that only the effects of the g component of M are considered. If two such volume increments are considered side by side as in Figure 3-9(c), then the bound sur- face currents along their common sides, with densities designated by Jon,y, cancel partially to produce a net upward flow of current in the region given by aM, de Aly = (Jang ~ Fong) Az = — Aw he ‘This current passing through the cross-sectional area Ax Az is depicted by the bold arrow in the figure. They component of the bound current density Jy, through Ax Az Av elements separated to show ayy, 294 Far, y) FIGURE 3-9. Relative to J, = V x M. (a) Bound current elements producing surface currents on Av. (6) Bound surface currents smoothed into rectangular components, assuming M, only. (c) Net volume current A/, through Ax Az: the difference of bound surface current densities. (¢) The other contribution to the J_,y component.4 MAGNETIC POLARIZATION AND GURL H FOR MATERIALS = 131 is then AI,/Ax Az = —6M,/éx. Another contribution, shown in Figure 3-9(d) is ob-. tained from the x component of M in the vicinity of the point; it contributes the density 6M,/éz through Ax Az. The totaly component of J,, therefore becomes J,,,y = @M,/0z — @M,/éx, which from (2-52) is evidently the y component of curl M. A similar develop ment yields the other components Fy. and Jn,z Of Jy, obtaining (3-56) a, ay a é@ 6 @ zB RTYM [3-36] M, M, M, The significance of (3-96) in revealing the presence of volume currents inside a material whenever its interior is nonuniformly magnetized is described in an example to follow. A side effect is the presence of surface current densities J,,, established by M on the surface of the material. EXAMPLE 3-3. Suppose a B field is applied to a cube of magnetic material, # m on a side, such that M is z-directed and varies linearly with x according to M = a,10x A/m, as shown in (a). Find the magnetization current density J,, in the material, as well as the surface magnetization current density. Sketch the bound current fields in and on the cube. ‘The magnetization current density Jp, is obtained from (3-56) a a, a 6 8 8 Jn=V XM = lo B® ayl0 A/m? a) 0 0 10s} negative y-directed and of constant density as in (6). ‘The uncanceled segments of the bound currents at the surface of the block constitute a surface density of magnetization currents denoted by Jyy (A/m). On the end x = by Jon is y-directed and has a magnitude equal to that of M there; that is, Jon)eay 7 2M 108 Alm ®) © surtace bound current flux EXAMPLE 3-3. (a) Material sample maguetized linearly with increasing x. (0) Volume mag- netization currents produced by transverse variations of M. (<) Surface currents produced by ‘uncanceled segments of bound currents.132 MaxwELL’s EQUATIONS AND BOUNDARY CONDITIONS while on the top and bottom of the black Solan = Jaheo™ ae hye = —a,10x A/m 3) 10x Afm (4) No bound currents exist on the end at x = 0, since M =0 there. These surface effects are shown as flux plots in (¢). ‘The modeling of the bound currents in Figure 3-9 reveals that, on any sampling cube Av, the surface bound-current densities Jj» are oriented perpendicularly with respect to the local M field. It is therefore evident that J, on any surface element of ‘Av can be found from the cross product of that M with the normal unit vector emerging from the surface. Thus, Jm= 02 xM (3-57) In Example 3-3, on the surface y= of the magnetized block, with m= a, and M =a,100 there, one obtains Jy, = —2 X M-= ~a, x a,10b = a,l0b, which agrees with the result (2) obtained in that example. The curl of B/jig in fice space has been expressed by (2-63) as the sum of a convection or a conduction current density J plus a displacement current density a(e,E)/Ai at any point. Two additional types of current densities occur generally in materials: J, = GP/dt of (3-35) and Jp, = V x M of (3-56), arising from dielectric and magnetic polarization effects, respectively. Adding these together accounts for the total current density at any point, yielding a revision of (2-63) for a material region. vx (2)-3+ eave Ho 7 a at Grouping the curl terms and the time-derivative terms together obtains Bog) yy HeoB +P) vx G ™) = (3-58a) Recalling from (3-23) that €9E + P defines D, and further abbreviating B/yg — M in (3-37) by use of the symbol H, sometimes called the magnetic intensity field = _—MAjm (3-58b) Ho permits writing (3-58a) in the compact form VxH=jJ+ (3-59)84 MAGNETIC POLARIZATION AND CURL H FOR MATERTAIS 133 This is the desired Maxwell curl expression for the field Hi defined by (3-58b), applicable to material regions, Note that it properly reduces to its free-space form (2-73) on setting P=M=0. In a Linear region possessing a magnetization M, one might be inclined to express M proportional to the B field in the material (i.c., M oc B) to provide a result analo- gous to (3-25) for a linear dielectric (P oc E), Historically, however, this has not proved to be the assumption used; instead, it is customary to set M proportional to Has follows: MaH = Xn (3-60) in which the dimensionless 7» is called the magnetic susceptibility of a material. Inserting (3-60) into (3-58b) therefore yields u=2_m=3_,0 Bo Ho which, on solving for B, obtains B= (1+ Zq)itoHl Wh/m? @-61) ‘The quantity (1 + %), abbreviated j1,, Hy = 1+ Yn (3-62) is called the relative permeability of the material. Further choosing the symbol p, called the permeability, to denote the product B= (1+ Xm)Ho (3-68a) H = tho Him (3-63b} permits writing (3-61) in the compact form for linear materials B= (1+ Xm)HoHl (3-64a) B= p,stoH (8-64b) B= yHWb/m? (3-64c) One should note the analogy of the steps yielding (3-64c) to those leading to (3-30), connecting D and E for linear, electrically polarized materials. It is scen from (3-64b) that the relative permeability expresses the permeability of a material relative to that of free space, Hg, if one writes “ — 3-65 Ho ‘ ? This is evidently analogous to (3-31), the expression for the relative permittivity ¢,.134 MAXWELL’s EQUATIONS AND BOUNDARY CONDITIONS A. Integral Ferm of Ampére’s Law for Materials “Maxwell’s curl relation (3-59), V x H = J + (@D/00} can be transformed into ar integral relationship by using Stokes’s theorem, Forming the dot product of (3-59) with ds and integrating over any surface 5 bounded by the closed line ¢ yields [vw de [yard [Da From Stokes’s theorem (2-56), the left side can be expressed as an integral of H+ dé over the closed line # bounding 5, assuming H suitably well-behaved; thus Gu dahy-ds+ 4 [Deda (3-66) the desired integral form of Maxwell’s differential equation (3-9). Equation (3-66) is also known as Ampere’s circuital lazo for materials. It states that the net circulation of H about any closed path ¢ is a measure of the sum of the conduction (or convection} current plus the displacement current through the surface § bounded by ¢. ‘Another curl relation, (3-56), J, = V * M connecting the magnetization field M with a volume magnetization current density, was treated in the last section. It has an integral form analogously obtainable by use of Stokes’s theorem, becoming fmeae= This means that the circulation of the M field about a closed path ¢ is a measure of the net magnetization current through it, For example, a surface integration of J, over a cross section in the 4-2 plane of the magnetized cube in Example 3-3 is seen to yield a bound magnetization current 105? A flowing vertically through the specimen, also obtainable from a line integral of M- df around a horizontal perimeter of the cube. -dsA (3-67) B. Boundary Conditions for Tangential H and M In a manner resembling the derivation of the boundary condition (3-41), one can compare the éangential components of H adjacent to an interface separating two materials, by applying Maxwell’s integral law (3-66) to the small, rectangular closed line £ shown in Figure 3-10, With the magnetic fields in the adjacent media labeled H, and Hz, and resolved into normal and tangential components as in Figure 3-10, integrating the left side of (3-66) clockwise around ¢ yields Hy A¢— Hi, Aé, if the height 6h is taken so small that the ends do not contribute to the line integral. ‘The right side of (3-66) involves integrations of J and D over the vanishing surface § bounded by ¢, obtaining aD, aD, Behe = ude bh + SP ALO (3-68) Hy Ad — Hyg Ad = JyAs + if 7, and D, denote the components normal to As. The last term of (3-68) vanishes as 6h “> 0; similarly, the contribution of the J, term would also vanish if J were a volume3-4 MAGNETIC POLARIZATION AND CURL H FOR MATERIALS = 135 Region 1 Ga) Surface current flux Region 2 (u,) FIGURE 3-10. Rectangular closed line ¢ constructed to compare H,; and H,, using Ampére’s law. current density. In some physical problems, however, one can assume a free suyfiace current flowing solely on the interface with a density J, defined by J. = lim Jon (3-69) oh00 (It develops that J, is of interest only if one of the regions is a perfect conductor, a case to be discussed shortly.) Thus, the general boundary condition resulting from the substi- tution of (3-69) into (3-68) becomes Fy — Ha = Jam Af. {3-70a) in which the subscript (x) denotes a surface current flowing normally through the side of the rectangle, as noted in Figure 3-10. Equation (3-70a) states that the dangential component of the FI field is discontinuous al an interface lo the extent of the surface current density that may be present. Using n to denote a normal unit vector directed from region 2 toward region 1 as in Figure 3-11, a vector form of (3-70a) is written a x (H, — H, JAfm (3-70b) to include direction as well as magnitude information, The boundary condition (3-70) is true in general, though in its application to a boundary-value problem, it becomes two cases.136 MAXWELL’ EQUATIONS AND BOUNDARY CONDITIONS Region 1: Gu, 6.00) Region (ug, e1609) Pe a St rent Region 2: WW ez surface density J, ee Region 2: (ug. €2. 099 i oli aes ie (a) (6) FIGURE 3-11. The two cases of the boundary condition (3-70b) on tangential /, (a) Continuous Ff, at interface separating regions of finite conductivities, (6) Equality of the H, and surface current density on a perfect conductor. Cast A. Both regions have finite conductivities. In this case, free surface cur rents cannot exist on the interface, reducing (3-70a) to Hy = Hey Sim (71) Thus, the tangential component of H1 is continuous at an interface separating two materials having, at most, finite conductivities. This boundary condition as illustrated in Figure 3-11{a}. Cast B. One region is a perfect conductor. From (3-44) it has been noted, under time-varying conditions, that no clectric or magnetic field can exist inside a perfect conductor. If region 2 were a perfect conductor, then H, = 0 reducing (3-70a) to Hy, = Fam; oF in vector form, (3-70b) becomes nx Hy =J,A/m (3-72) the boundary condition depicted in Figure 3-11(b). At the interface separating a region from a perfect conductor, the surface current density J, has a magnitude equal to that of the tan- ‘gential HL there, and a direction specified by the right-hand rule. It is shown later that no normal component of H or B may exist at the surface of a perfect conductor, implying that the tangential magnetic field is also the tofal magnetic field there. ‘A similarity in form is noted between Ampére’s circuital law (3-66) and the relationship (3-67) for M. Thus, by analogy with the boundary condition (3-70a), derived by applying (3-66) to the closed rectangle as in’ Figure 3-10, one may establish from (3-67) the boundary condition Ma ~ Mex = Fam Aim (3-73a) "This result expresses the continuity of the tangential component of M as one traverses an interface between two adjacent, magnetized regions. The subscript (x) denotes a24 MAGNETIC POLARIZATION AND CURL H FOR MATERIALS 137 surface magnetization current density normal to the tangential M components at the boundary. The vector sense of the surface magnetization current density Jon is included in the boundary condition (3-73a) by expressing it mx (M; —M)) = Jon Af (3-73b) a result analogous with (3-70b). Ifregion 1 is nonmagnetic, then My = 0, reducing (3-73b) to (3-37) Jon = = XM, A/n (3-74) An illustration of the latter has already been noted in parts (6) and (c} of the figure accompanying Example 3-3. EXAMPLE 3-4. Suppose a very long solenoid like that of Figure 1-21{b) contains a coaxial magnetic rod of radius a, as in figure (a), the rod having a constant permeability 1. The winding is closely spaced with x turns in every dm of axial length, carrying a steady Magnetizable 26865808985 S5SSSSSESES SH9080008 (b) (e) @ ‘Grossly enlarged FLUX Of Tam bound currents (e) EXAMPLE 3-4. (a) Solenoid with magnetic core. (6) Held flux. (¢) B field flux. (@) ML field flix. (2) The Jy field on the iron.138 sxweLt’s EQUATIONS AND BOUNDARY CONDITIONS current [, (a) Determine Hi and B in the air and iron regions, making use of Ampire’s law (3-66) and symmetry. (b) Find M in the rod, and determine whether any volume magnetization current density J,, exists in it, as well as magnetization current densities on its surface. (¢) Sketch the flux of H, B, and M in the air and iron regions. (a) With de in the wire producing time-s ) tic fields, (3-66) becomes §,H- df = [, J+ ds. From the axial symmetry and the implications of Ampére’s law in relation to the current sense, H is positive z directed within the winding and essentially zero outside it, Constructing the rectangular closed path ¢ showa, H+ dé integrated between P; and P, yields GH de = JP (aH) ade = fp Hide = nl in which #1, is constant over the path P, to Pa, yielding 09) This result is correct in both the air and iron regions because n/ is the current enclosed by / regardless of whether P, and P, fall within the air or the iron. The turns per meter in the winding are denoted by nfd. ‘The corresponding B field is obtained from (3-64c} B= pH pla,tt) = 0! 0 wo) (b) EXAMPLE $-5. (a) B flux refraction. (3) Refraction at air-to-magnetic-region interfaces. nents are (3-50) and (3-71); By = Byg and H,, = Hy, The latter can be written, Pa Pa (3-75) A He From the geometry of the figure, the tilt-angles obey tan 6, = Bj,/B,, and tan 0; = Ba/Bya, which combine with (3-75) to yield et 1 tan 8, = 5 Inserting the expression for tan 8, obtains tan 0, ="? tan @, (8-76) on140) MaxWELL’s EQUATIONS AND BOUNDARY CONDITIONS ‘Asa numerical example, compare the tlt of the B lines at an interface separating two re- gions with 4 = Hg and ff, = 10}éo. Assume at some point on the interface that B in region 1 is tilted by 0, = 45°, From (3-76), 0, = arc tan (10 tan 45°) = 84.3°, Similarly, if @, = 20°, then 8, = 74.6”, and so on. In the event of an air-to-iron interface (ftz > #44), one may show from (3-76) that for nearly all 0, the corresponding @, vaiues are very small angles (essentially 0°); that is, the flux leaves the iron nearly perpendicularly from its sur- face. These examples are noted in (6). C. The Nature of Magnetic Materials The classical macroscopic theory of the field phenomena associated with mag- netizable substances, introduced in Section 3-4, attributes their magnetic properties to the magnetic moment mi provided by the orbiting electrons, electron spins, and nuclear spins. Moreover M denotes from (3-55) the averaged volume contributions of the magnetic moments m in the vicinity of any point inside the substance, The net magnetic effects are altered significantly by the temperature—the random thermal agitations that inhibit the alignment of the magnetic moments. Although noteworthy advances in the understanding of magnetic processes on the microscopic scale have been provided by applying quantum mechanics and electromagnetic theory to models of the magnetic elements, there is yet much speculation in the deduction of the magnetic properties of the many complex alloys and compounds. Magnetic effects in materials have been classified as diamagnetic, paramagnetic, ferromagnetic, antiferromagnetic, and ferrimagnetic. The following discussion is in- tended to provide a glimpse of some of the classical models of magnetism to help explain the origins of these magnetic properties.* In a diamagnetic material, the net magnetic moment m of each atom or molecule is zero in the absence of an applied magnetic field. In this state, the classical picture of the electron speeding at an angular velocity @ about a positive nucleus is accompanied by a balance of the centrifugal and the attractive Coulomb forces between those op- posite charges, The application of a magnetic field provides a Lorentz force, —ev x B, on the orbiting electron such that ifa fixed orbit is to be maintained, an increase or de- crease +Aq@ in the electron angular velocity must occur, depending on the direction of the applied B field relative to the orbital plane. This amounts to a change in the electronic orbital current, thereby generating a small magnetic field, the direction of which is such as to oppose the applicd ficld. The net, opposing magnetization field M thus created in any typical volume-element Ao of the material leads to a slightly negative susceptibility ,, for such a material. Diamagnetism is presumed to exist in all materials, though in some it may be masked by other magnetic effects to be discussed. Typical small, negative values of %, for diamagnetic solids at room temperature are 1.66 x 1075 for bismuth, —0.95 x 1075 for copper, and —0.8 x 10-5 for germa- nium. It is to be expected that the less dense gases have even smaller diamagnetic susceptibilities, which is borne out by both calculation and experiment. ‘Another weak form of magnetism is known as paramagnetism. In a paramagnetic material, the atoms or molecules possess permanent magnetic moments due primarily to electron-spin dipole moments, randomly oriented so that the net magnetization M of (3-55) is zero in the absence of an applied magnetic field. The application of a B field to gaseous, paramagnetic nitrogen, for example, produces a tendency for the moments m to align with the field, a process inhibited by the collisions or interactions among the particles. In a paramagnetic solid, thermal vibrations within the molecular 5An excellent digest of the theories of magnetic phenomena, including ample references, is found in Chap- ter 7 of R.S. Elliott, Electromagnetics. New York: McGraw-Hill, 1966.34 MAGNETIC POLARIZATION AND CURL H FOR MaTERIATS 141 Transttion region No Weak ~— Moderate Saturation, applied’ applied field field field field © @ FIGURE 8-12, Alignment of magnetic moments in a ferromagnetic material: domain phenomena. (4) Magnetic moment alignments in a ferromagnetic material. {6) A perfect single crystal, showing domains and domain walls. (o) A transition region between adjacent, domains. (d) Domain changes in a crystal with increase in the applied B field. lattice tend to lessen the alignment effects of an applied magnetic field. The room temperature suscoptibilitics of typical paramagnetic salts such as FeSO,, NiSO,, Fe,O, and CrCl, are of the order of 107 and inversely temperature-dependent, according to a law discovered by Pierre Curie in 1895. The important ferromagnetic materials are characterized by their strong, perma- nent magnetic moments, even in the absence of an applied B field. They include iron, cobalt, nickel, the rare earths gadolinium and dysprosium, plus a number of their al- loys and even some compounds not containing ferromagnetic elements. It was originally postulated by Weiss in 1907, and much later confirmed experimentally in photomicto- graphs by Bitter,® that a ferromagnetic material in an overall unmagnetized state in reality consists of many small, essentially folally magnetized domains, randomly oriented to cancel out the net magnetic field. Domain sizes have been found to range from a few microns to perhaps a millimeter across for many ferromagnetic materials. Weiss further postulated that strong intrinsic coupling or interaction forces exist between adjacent atoms to provide the fully magnetized state within a given domain. It was not until 1928 that Heisenberg of Germany and Frenkel of the U.S.S.R. independently verified, using quantum theory, that the extraordinarily strong forces holding the domain atoms in parallel alignment is attributable to the coupling forces between the net electron spins of the adjacent atoms.” The parallel orientation of the spin moments in a ferromagnetic domain is depicted in Figure 3-12(a). An idealized, perfect crystal might have a domain structure, in the absence of an applied B field, like that shown in Figure 3-12(6), although flaws such as lattice imperfections and impurities would modify this idealized picture somewhat, The walls between the domains (Bloch walls), having the appearance suggested by Figure 3-12(c), are transition regions between - the spin alignments of the adjacent domains, and they are of the order of 100 atoms ©F, Bitter, “A generalization of the theory of ferromagnetism," Pips. Rev., 54, 79, 1938. TW. Heisenberg, “On the theory of ferromagnetism,” Zeit, f. Phys., 49, 619, 1928.142 — MaxweELi’s EQUATIONS AND BOUNDARY CONDITIONS Magnetization M Irreversible - magnetization rotation region, Irreversible wall motion region Applied H Sinusoidal | applied H field fe (bh) FIGURE 3-13. Magnetization effects due to an external magnetic field applied to a ferro- magnetic material. (a) Magnetization process (solid line) in a virgin ferromagnetic region. Irreversible behavior shown dashed. (8) B — Hf hysteresis loops for a ferromagnetic material. thick. The domain division by such wall structures occurs in such a way that a minimal external magnetic field is supported by the structure, to minimize the work done in forming the structure. , Asan external B ficld is increasingly applied to a ferromagnetic crystal containing domains, as denoted in Figure 3-12(d), the Bloch walls first move to favor the growth of those domains having magnetic moments aligned with the applied field, a reversible condition on removing the field if B is not too large. For higher applied fields, domain- wall motion occurs, which is not reversible, as noted in the third sketch of Figure 3-12(d). For a sufficiently large applied field, the domain magnetic moments rotate until an essentially total parallel alignment with the applied field occurs, a condition called saturation. The averaged effect of such changes on the bulk magnetization M, in a sample volume clement containing a sufficient number of domains, is shown in Figure 3-13(a). The arrows denote the direction of increasing or decreasing the applied H field.® One may note, on decreasing the applied H field to zero from the values at P; or P3, that a permanent magnetization M,; (or M,2) is retained in the ferromagnetic sample, signifying an irreversible and distinctly nonlinear, multivalued behavior. These M, values-are termed the remanent (remaining) magnetizations of the specimen. The applied field must be further decreased to the reverse value H,, (ot H,,) as shown, before *It has become customary to denote the applied magnetic field in the material as the H field, rather than B.3-4 MAGNETIC POLARIZATION AND CURL H FOR MATERIALS = 143, the permanent magnetism is removed from the material. The value H, is rather loosely called the coercive free, the field required to reduce the magnetization to zero within a specimen. If the M-H plot of Figure 3-13(a) is replotted in terms of the B field in the ferro- magnetic material, the B-H curve of Figure 3-13(5) results, recalling from (3-58) that these quantities are related to B = ji(Ei + M). In (6) is depicted a complete cycle of the events of (a), such as might occur if the applied field were varied sinusoidally as noted below the B-H curve. After the virgin magnetization excursion from 0 to P3, obtained over the first quarter cycle of the sinusoidal H field, the subsequent decrease in H provides the sequence of values passing through the remanent value B,, the co- ercive force H,, and thence to the maximum negative flux density in the material at P,. With the applicd ficld going positive once more, a reversed image of the prior events takes place. The multivalued curve obtained in this cyclic fashion is called the Iysteresis (meaning lagging) loop of the ferromagnetic region. Note that for smaller am- plitudes of the applied H field, correspondingly smaller hysteresis loops are obtained, whether centered about origin 0 as just described, or appearing about Po as the con- sequence of a bias field Ho. The incremental permeability of a ferromagnetic material is defined as the slope of the B-H curve. The slope at the origin 0 of the virgin curve is called the initial incre- mental permeability, If the material is used such that it possesses a fixed (de) magneti- zation Hg with a small sinusoidal variation about this value as noted at the point Po in Figure 3-13(6), the minor hysteresis loop formed there has an average slope defining the incremental permeability there. These events take place in the ferromagnetic core of an inductor or transformer coil carrying an alternating current superimposed on a direct current, for example. Energy must be expended in supplying the losses incurred in the hysteresis effects accompanying the sinusoidal variations of an applied field. For this reason, ferromagnetic materials with low coercive forces (having a thin B-H loop) are desirable for transformer and inductor designs. On the other hand, a ferro- magnetic material used for permanent magnets should have a high coercive force H, and a high remanent, or residual, flux density B, (corresponding to a fat B-H loop). ‘Table 3-1 lists a few representative ferromagnetic alloys along with some of their magnetic properties. ‘An additional and usually undesirable side effect, occurring in the magnetic core of devices such as transformers, is that of the free-electron conduction currents circulating within the core material due to an electric field E generated inside it by a time-varying magnetic field. The densities of these currents are limited by the con- ductivity of the core material through (3-7), that is, J — E, and are given the name eddy currents because of their vortexlike nature within the conductive core, resulting from their relationship to the time-varying B field through (2-62) B VxE=—— [2-62] In the next section (2-62) is shown to be valid for a material region as well as for free space. Thus, with a conductive, ferromagnetic core in the solenoid as shown in Fig- ure 3-14(a), a sinusoidally time-varying current in the winding produces a sinusoidal B ficld in the core material to generate an E field, and from (3-7) also an eddy current field therein. Its sense is thus normal to the time-varying B field, ‘The losses may be reduced substantially by subdividing the conductive core into a fibrous or laminar struc- ture, as suggested by Figure 3-14(0), in which the subdivided conductors are insulated from one another. Small, spherical magnetizable particles serve the same purpose.ares ™D & So" 00°F ‘00 8 IN FL TVS A oorapy 1 00°F 2g 196 60 UN T [ais voqery (Guufaiv “# INJNVINS (WN) BONO! AAINIOO NOWULSOdWOS INSOYd WRAIVIN ‘Symaaoue youSous juounussag (a) sro ~ opoere £0 o00%008 000°00% au OT “One S IN OL Aoreuuadng aro ¥ LoL oo0%oot 000°6 ad #16 “OW 9'0 IN BL AoTpeusad gL zo or z 00's 00s‘ 3a 96" (poauapio uyesB) psxodApy 910 OF a a00% 00% 2496 IS F wosT WOOTTS, (aya 0b *) (ui) Guia) WWNINIXY TILNI NOUISOdNOO INIDREd TREY NIALONGNOD: "H 3ONOY JAINNIOO 8 NOUNS: Sonor aT ‘shoppe aausofeuoay (Vy) shojty oWeUBDWOUE, JO Seyedolg OYEUBDIN +e TEL x * =34 MAGNETIC POLARIZATION AND GURL H FOR MATERIALS = 145 Solenoid winding. RARAHERRAE Conductive, magnetic core Band d flux (induced eddy currents} (a) (by FIGURE 3-14, Eddy currents im conductors immersed in time-varying B fields. (a) Eddy cur- rents induced in a conductive magnetic core by 2 time-varying B field. (¢) Fibrous and laminar core structures used to break up eddy current paths. This constrains the eddy currents to much smaller volumes, limiting their densities substantially if the cellular substructures are made sufficiently smail or thin. In the previous discussions it was seen that paramagnetism is a characteristic of materials possessing permanent magnetic spin moments that are randomly oriented, a condition depicted in Figure 3-15(a}. Ferromagnetic materials, due to the effects of short-range couplings between adjacent atoms, possess parallel-oriented atomic mag- nets within given domain boundaries that comprise the material as suggested in Figure 3-15(6). If such a material is heated until the thermal energies exceed the coupling energies, the material becomes disorganized into a paramagnet, though on cooling it reverts to a ferromagnet once more. The critical temperature at which this occurs is known as the Curie temperature. Variations of the coupling phenomena responsible for ferromagnetic materials can even produce antiparallel alignments of electron spins in materials known as anti- ferromagnetic, as depicted by Figure 3-15(c). In this state, an antiferromagnet is charac terized by a zero magnetic field. Manganese fluoride, for example, is paramagnetic at ak VE 7 “Nal fttttate yt tay ty (a) () (cy tte that ay fitetety (a) FIGURE 3-15. Orientations of the spin moments of various magnetic ma- terials, (a) Paramagnetic. (}) Ferromagnetic. {c) Antiferromagnetic. (d) Ferti- magnetic materials, or ferrites.146 maxweri's EQUATIONS AND BOUNDARY CONDITIONS room temperature, but on cooling it to ~206°C: (called its Néel temperature, after the French physicist}, it becomes antiferromagnetic; below this temperature it ex- hibits no magnetic effect. An important variation of this phenomenon is ferrimagne- tism, associated with noncanceling antiparallel arrangements of the coupled spin moments as suggested by Figure 3-15(d). Thus, in magnetite, the magnetic iron oxide FeO - Fe,Os, two of the three adjacent spins are reversed such that a somewhat weaker form of ferromagnetism is produced. Magnetite is an example of the group of ferri- magnetic oxides XO - Fe,Os, in which the symbol X denotes a divalent metallic ion Ga, Co, Cu, Mg, Ma, Ni, Zn, or divalent iron. When synthesized in the laboratory, these brittle, ceramiclike compounds are particularly useful for magnetic cores in high- frequency transformers and special applications ranging into the microwave frequen- cies because of their low conductivities comparable to those of the semiconductors, usually from 10-1 to 10~* O/m. They are thus desirable because they limit eddy current losses in such applications. ‘These conductivities may be compared with the much higher values of the typical alloys for lower-frequency applications as listed in Table 3-1, in which the values of the order of 10° G/m appear. A general account of the theory of ferro- and ferrimagnetism, together with a number of microwave appli- cations of the latter, is found in the book by Lax and Button.? 3-5 MAXWELL’S CURL E RELATION: ITS INTEGRAL FORM AND BOUNDARY CONDITION FOR TANGENTIAL E In Section 3-3, the Maxwell relation (3-59) for eurl Hin a material region was developed by adding in those current densities contributed by the electric and mag- netic polarization fields. The form of the eurl E relationship for materials is obtained by analogy, but retaining its form (2-62) for free space B VxE=—>- (3-77) That the free-space Faraday’s law (2-62) remains correct for a material region is evident on observing that an additive magnetic-current-density term, analogous to the. clectric-current-density term J of (3-7), is physically impossible if free magnetic charges cannot exist. Thus (3-77) correctly applies to both materials and free space. Equation (3-77) is readily converted to an integral form. The scalar multiplica- tion of (3-77) with ds, integrating the result over any surface § bounded by a closed line ¢, and applying Stokes’s theorem yields Eds 4 fipeusy (3-78) again unchanged from the free-space version (1-55). ‘The determination of (3-77) and (3-78) completes the development of Maxwell’s differential and integral relations applicable to material regions, and they are sum- marized in the first two columns of Table 3-2. "B, Lax, and K. J. Button, Microwave Ferrites and Ferrimagnetics, New York: McGraw-Hill, 1962.y , i (6-6) 0=@q—'a) xm 10 tg ay {92-¢] soa) eR -p-a$ leel a7 =axA fees] f="Hxe fies] H="H co < to 2g 0) onuy Fo To sy a9, i 4 i T= CHW xe 0 Lay Oy woe) vaf Zee f=pn§ — (osel Zttquxs (ove) = (Ca—ta) = 0 Mg= My [ore] o=#-a'f [9-6] O=8-A (sel ‘="a tare) "q="¢q we to cg a0 ouaz 29 ‘19 27 9907) ‘d= (q-'q)-u wo d= — Mg es-é] wdt=s-ah bes) "d= ak NOILKINOS AuWONNOP ONIGNOdsRCS- WaOd WeaiNi WdaOd TWUNSU32I0 SO_YE4U] UD JD SUOLIPUED AIpUNog JOYDAS Bulpuodseuo ey} pud suoyonby syjemxoy| ADWUWNS 7+ SIEVL 147148 — saxwel.t's EQUATIONS AND BOUNDARY CONDITIONS ‘A boundary condition, comparing the tangential components of the E fields to cither side of an interface, may be obtained from Faraday’s integral law (3-78). The details of the derivation may be avoided if one recalls that Ampére’s line-integral law (8-66) leads to the boundary condition (3-70a), Hy: — Hr2 = Jam" The boundary con- dition comparing the tangential components of E can be analogously found by apply- ing (3-78) to a similar thin rectangle, yielding the analog of (3-70a) Eu — Ea (3-79) ‘Thus the tangential component of the E field is always continuous at an interface. The right side of (3-79) is evidently zero because no magnetic currents are physically possible. ‘A summary of the four boundary conditions derived from Maxwell’s integral laws for material regions in Sections 3-2C, 3-4B, and in the present section, is given in Table 3-2. EXAMPLE 3-6. (0) Derive a refractive law for E at an interface separating two nonconduetive regions. (b) Deduce from’ boundary conditions the direction of E just outside a perfect conductor (a) ‘The boundary conditions for the tangential and the normal components of E at an interface separating nonconductive regions are (3-43) and (3-79); that is, Ey, = €:F ga and Ey = E,,. From the latter and the geometry of (a), one obtains tan 0, = 2 tan 9 (3-80) 1 a result analogous with (3-76) of Example 3-5 concerned with the reftaction of B lines. {B) From (3-44}, a perfectly conductive region 2 implies null fields inside it. Then by (3-79). Ey mn the adjacent region | must vanish also. The remaining normal com- EpenEq ponent in region 1 is given by (3-45), Dui = Ps, Yielding p, = € By as shown in (6). | fe = 61 En ie , OE LEN (b) (a) “ EXAMPLE 3-6. (2) E flux refraction at an interface separating nonconductive regions. (3) E is everywhere nermal to the surface of a perfect conductor. Region 1: (1. €1, 91)3-6 MAXWELL'S CURL E RELATION: ITS INTEGRAL FORM 149 Surface current (a) (b) EXAMPLE 3-7. (a) Parallel-plate system supporting a uniform plane wave field. (6) Charge and current diseribution on conductor inner surfaces. EXAMPLE 3-7. A uniform plane wave is described by the electric and magnetic fields E=a,lz(z, f) = a,Fm cos (ot — Boz) + == 08 (Wt — Boz) H= aH} (2,0 = No and propagates in air between two perfectly conducting, parallel plates of great extent, as in (a). The inner surfaces of the plates are located at ¥ = 0 and x = a, Obtain expres- sions for (a) the surface charge field and (b) the surface currents on the two conductors. (a) The given E is everywhere normal to the plates at x= 0 and « = a, satisfying the boundary condition of (4) in Example 3-6, The surface charge distributions thus become Ea, AEs = egLin cos (Wt — Boz) po=m'-D, =. ea, - a, BS = —egBe cos (et — Boz) pp= aD, = con E; =0 a implying that E lines emerge from positive charges and terminate on negative ones. {) The given H, to satisfy (3-72), must be everywhere tangential to the perfect con- ductors at x = 0 and x = a, yielding there 1 XH =a, FE co (oF ~ fo =nxH, io = cos (ot ~ Bos) #4 ja Jona! xi = a, x aiff = It is seen that, in any fixed z plane, current flows in opposite z directions in the two conductors.; | | 150 axwet’s EQUATIONS AND BOUNDARY CONDITIONS 3-6 CONSERVATION OF ELECTRIC CHARGE A relationship between charge and current densities is obtainable from Maxwell’s equations, assuming that electric charge can neither be created nor destroyed. Let a charge density py(u4, uz, us, f) occupy some volume region V. Then the net charge in V at any instant is \ 419 = feels, us ts, Ode © Note that even though p, is in general a function of both space and time, the net q enclosed is a function of ¢ only, because the definite limits on the integral dispose of the space variables. For brevity, the latter is written with the function notation understood as follows. a= fede (3-81a) ‘The time rate of change of q within V is a measure of the current flowing into the closed surface § bounding V; hence aq ¢ op, Gf Peng, 3-810) + f FE dy Cfsce or A (3-81b) With ds directed normally outward from S, the current flowing out of S' becomes Jrds (3-82a) implying that the net positive charge q inside V is decreasing in time. The postulate that electric charge is neither created nor destroyed permits equating the negative of (3-81b) to (3-82a}, yielding QJ d= ~ fea (3-82b) This means that the net outflow of current from any volume region is a measure of the time rate of decrease of electric charge inside the volume. Equation {3-82b) is thus the expression of the conservation of electric charge. ‘The relation (3-82b) has an equivalent differential, or point form obs Vey= ét Ajm? (3-82c) a result obtained by applying (3-82b) to any limiting volume-clement and using the definition (2-20) of divergence. While (3-82c) is truc for any volume-element of a current-carrying region, it is also applicable to the surface currents and charges at the interface between a perfect conductor and a perfect insulator, as in the system of the forthcoming Example 3-8. With currents and charges confined to the interface so that J > J, and p, — p,, the3-6 CONSERVATION OF ELECTRIC CHARGE 151 charge-conservation relation (3-82c) becomes!® 6 Vee Je= ~Et Alm? (3-82¢) if Vz + J, is taken to mean a tangential (two-dimensional) surface divergence of J,. For example, if the interface coincides with the »-< plane, implying J, = a, Jy +, the two-dimensional divergence of J, is written ya obey , Ose Vords dy tae In a time-static field problem, steady current densities are divergenceless, so (3-82c) reduces in that case to v-J=0 (3-82e) Direct currents are therefore always characterized by uninterrupted, closed current flux lines, EXAMPLE 3-8. Show that the surface current and surface charge fields at the conductor di- electric interfaces of Example 3-7 satisfy the two-dimensional charge-conservation relation (3-82d). ‘At the lower interface (at x = 0), the left side of (3-82d) yields an, Bo vp ja T'S e to Eq sin (@t — Boz) = +0 og sin (wt — foe) on substituting By = @a/upéy and Mo = V/to/€o. With a surface charge density p, = + €9£q cos (wt — Boz) on the lower conductor, 4 = FE = + weplit sin (wt — By2) whence (3-82d) is satisfied. EXAMPLE 3-9, Determine the relaxation expression for the time decay of a charge distribution in a conductor, if the initial distribution at t= 0 is polity, uo, Us, 0). ‘The desired result is obtained by combining (3-82c) with the expression for div D. Replacing J with oE for the conductive region obtains, ftom (3-82c) OR. V+ (0B) + ay The region being homogeneous makes € and ¢ constants, so (3-24) is written V-E = pyle, and substituting it into the first term of (1) yields ép, teh (2) ‘It should be noted that the relationship (3-82c), connecting current density and charge density at any point in a region, is consistent with the Maxwell curf expression (3-59). This is evident from taking the divergence of the latter, which promptly yields (3-82c) on making use of the identity (19) in Table 2-2,452 MAXWELL’s EQUATIONS AND BOUNDARY CONDITIONS Integrating yields the desired result Protas Has tgs 4) = Pools, Way a, Oe! assuming the initial charge distribution at t= 0 to be Pyolts, to, ts, 0). This if the internal, free electric charge in a conducting region is zero, it will remain zero for all subsequent time. One may conclude that in a material having a nonzero conductivity «, there can be no permanent volume distribution of free charge. Thus, the siatic state of a free charge supplied to a conducting body is that it must ultimately reside on the surface of the conducting body through the mutually repulsive (Coulomb) forces among the free charges ‘The time constant t of the free charge density decay process (3-83a) in Example 3-9 is given by (3-83b) a quantity called the relaxation time of the conductor. Good conductors, for which ¢ may be of the order of 107 U/m, have relaxation times around 10778 sec, assuming a permittivity essentially that of free space. In poor conductors, 7 may be of the order of microseconds, though a good insulator may have a relaxation time of hours or even days. *3-7 UNIFORM PLANE WAVES IN AN UNBOUNDED CONDUCTIVE REGION The topic of uniform plane waves propagating in empty space was discussed in Sec- tion 2-10, in which the influence of the free-space parameters féy and €9 on the various wave characteristics was observed. The study of a plane wave propagating in a mate- rial having the parameters €, jt, and ¢ is considered in this section, It is shown- that the important new effects produced by the conductivity ¢ is to provide wave decay in the direction of propagation, as well as a phase shift between E and H. The assumptions made for the problem of wave propagation in an unbounded, linear, conductive region are 1, The components of E and H have neither x nor y dependence; that is, @/@x = a/éy = 0 for all field components, 2, Free-charge densities p, in the conductive region are in general nonzero if the charge-continuity relation (3-82c) is to be satisfied; while the current density J in the conductor!» is related to the E field therein by (3-7), J = oB. 3. The parameters of the region, assumed linear, homogeneous, and isotropic, are H, € and o. ‘The problem will employ time-harmonic forms of the fields, With 6, = 0 and J= cB, Maxwell’s equations for the region are obtained from (3-24), (3-48), (3-59), As an option, this section may be omitted for now, ta be taken up (along with Section 2-10) before be- ginning Chapter 6, ifdesired. However, its relevance to an improved understanding of the material param- eters over the broad frequency spectrum makes it desirable for study in this chapter. *1Although this assumption refers explicitly to waves im a conductive region, the extension to wave prop- agation ia a lossy dielectric through the use of a loss tangmt, €”/€, is described in Section 8-8.27 UNIFORM PLANE WAVES IN AN UNBOUNDED CONDUCTIVE REGION 153 and (3-77), becoming V- (cB) =p, (3-84a) v-B=0 . (3-84b) Vx B= —joB = —joult (3-84c) Vx H=J+joD = cB + jock (3-844) in which B = pH and B = eB of (3-30) and (3-64) are applicable. ‘These equations necd not in fact be solved, since this has already been done analogously in Section 2.10 for plane waves in empty space. To obtain the sohation by analogy, compare (3-84a) through (3-84d) with (2-106) through (2-109) applicable to the empty-space case V- (é9E) =0 [2-106] Vv-B=0 [2-107] Vx B= —jon fi [2-108] Vx A= joey [2-109] in which B = oH and D = eoE apply. A comparison of the two curl expressions in these two groups of Maxwell’s equations reveals that the two V x E expressions are precise analogs of each other, with (3-84c) obtainable from (2-108) on simply replacing Ho in (2-108) with x. Comparing (3-84d) with (2-109), however, reveal an additional conduction-current-density term GB in (3-84d). On collecting terms of the right side of (3-84d) as follows V x i= of + jocB = (6 + jog =i0(¢ -i2)e (3-85) the analogy of the latter with (2-109) is evident on replacing €, of (2-109) with the complex permittivity, € — jaf. Thus, each of the Maxwell’s equations (2-108) and (2-109) is seen to become (3-84c) and (3-84d) on replacing in the former Mo with wand with (« - iz) (3-86) ‘These replacements applied to the wave solutions of (2-108) and (2-109} are therefore expected to yield the solutions of (3-84c) and (3-84d) in an unbounded conductive region, Recalling the solution (2-115) for empty space E= EQ + Kw = Bye devise 4B ptaviien= [2-115]154 Maxweits EQUATIONS AND BOUNDARY CONDITIONS the replacements (3-86) in the latter yield analogous plane wave solutions for an unbounded conductive region B= Ei) + B® = Rigo Ove KOION 4. fF glo ele— Keron (3-87) In (3-87), the pure phase factor jor/pigég of (2-115) becomes a complex factor abbre- viated with the symbol }, called the propagation constant y=je u(« (3-88) and 7 can be separated into real and imaginary parts ~ =a+jpm7* (8-89) in which a, the real part of 7, is called the attenuation constant, and f is termed the phase constant of the uniform plane waves (3-87). Explicit expressions for a and B are found by replacing 7 of (3-88) with a + j8, squaring both sides to remove the radical, and equating the real and imaginary parts of the result. ‘The following positive, real solu- tions for a and f are obtained. Jue [ 1+ (g) . if Np/m (3-90a} V2 7 7 ‘ oe) A +(Z) “if! rad/m (8-90b) The dimension of # and f is (m)~1, though the artificial dimensionless terms neper and radian are usually mentioned to emphasize their attenuative and phase meanings in the wave expressions. With the substitution of (3-88) into the exponent of the wave solution (3-87), one may express it B= + B®) 8-91) = Ber + Bet (3-91b} = (Bib alO* \e-9%¢-H82 4. (Hig gl) ateib= (3-91c) in which the complex amplitudes £¥ of the traveling wave terms are denoted again as in (2-116) ER=skie Bo = Ej (3-92) A comparison of the conductive region wave solution (3-91) with the empty-space wave solution (2-115) reveals the presence of two real factors, e “ and ¢, accounting for wave decay-as the positive z and the negative z traveling waves proceed in their corresponding directions of flight with inereasing time. An additional view of the decay (attenuation) property of the waves is gained by converting (3-91) to its real-time form,