Sunteți pe pagina 1din 8

Is B or H the fundamental magnetic field ?

Johannes Hendrik Cloete Department of Electrical and Electronic Engineering University of Stellenbosch, Stellenbosch 7600 South Africa emaii: jhcloete@firga.sun.ac.za
1. So Maxwell had four field vectors - E, D, B and H the D and H were hidden ways of not paying attention to what was going on i n the material.
curl H - a D / d t divD

div B People tended to think that H was the magnetic field. But, as we h,ave seen, B and E are physically the Two of these, the laws of Faraday and the divergence law fundamental fields, and H is a derived idea. for B are homogeneous, thus source free. The other two, R. P. Feynman, The Feynman Lectures on Physics. the laws of Ampkre and Gauss are in general not source [l, pp. II-32-4 and 11-36-12] free. The free charge density, pfree(C/m3), in Gausss law. and the free current density, Jfree (A/m), of Amptres 2 . B y comparing the interaction product quantities with law, are related by the law of charge conservation divJf,,,+ the corresponding quantities derived from th,e Lorent:force laws and the laws of mechanics, we shall con- dpfree/dt = 0. clude that the E-H formulation is the correct formuT h e charge and current densities are free in the sense that lation of macroscopic electrodynamics. they are associated with the movement of charge carrierR. M. Fano, L. J . Chu and R. B. Adler, Electromag- s which are not bound t o stationary atoms or molecules netic Fields, Energy and Forces, [2, p. 4861 [4,pp. 172-3, 257-8, 309-111. T h e free current density in linear, isotropic matter can often be modelled as a diffu3. Whats i n a name 1 That which we call a rose sion current which satisfies Ohms law according to the B y any other word would smell as sweet. constitutive relation Jfree uE, with U the conductivi= William Shakespeare] Romeo and Juliet, II.2.43-44 ty of the medium and E the electric field. For unipolar convection currents like a n electron, proton or ion beam 4. To be or not t o be: that is the question. Jfree lpfreeIpEis a good model, with U the mobility of = , William Shakespeare, Hamlet, Prince of Denmark, the charged particles 15, p. 2521. Act 111, Scene I This paper is concerned with the meaning of the field vectors E and D, which are associated with electricity, and B and HI which are associated with magnetism, in the 1 Introduction Maxwell equations stated above. Circuit theory based on Kirchhoffs laws is not powerful enough for the purposes of many engineers. They must understand] model and analyze electromagnetic phenomena a t the level of Maxwells theory using quasistatic or fully dynamic formulations. Problems which require Maxtvellian models arise in the design of electric and magnetic machines; high voltage engineering; radio frequency and microwave engineering; antenna design; propagation and scattering predictions; and the design of dielectric, ferroelectric, ferrimagnetic, ferromagnetic, semiconductor and superconductor materials and devices. Maxwells differential equations for the electrodynaiiiics of matter, e.g. [3, pp. 126-1271 and [4, p. 3111, may be written in the form
For linear, isotropic, electrically polarizable matter the electricity vectors E and D are often linked by the constitutive relation D = [E and it may be tempting to conclude that E and D represent the same electrical phenomenon with the only difference being the factor E [4, pp. 1741751. However E, as defined in Section 2, is clearly the electric field, and as shown in Section 3 the D field is a secondary or auxiliary field which may be useful when dealing with dielectrics. There now seems t o be unanimous agreement amongst both engineers and physicists about the primary status of E and the secondary status of D [6, p. 2711.
N o t so in the case of t h e B and H fields which a r e used in the description of magnetic phenomena. Here confusion and disagreement still reigns, certainly among undergraduate students hut even among eminent and experienced engineers and physicists. There seem to be two camps [7]. In the first. where physicists are in the majority, it

= Jfree (Amp&reslaw) = pfree (Gausss law) = 0 (No magnetic monopoles)

curl E

+dB/dt

= 0

(Faradays law)

0-7803-3019-6/96/53.000 1996 IEEE.

3 54

is taken as axiomatic from the microscopic Lorentz force electromagnetic theory by allowing us t o predict the molaw that B is the primary magnetic field; then H is de- tion of charged particles such as electrons, and it gives a rived as an auxiliary field which is useful in the description unique definition of the electric and magnetic fields, a,nd of phenomena involving magnetically polarizable matter. hence the symbols E and B. in terms of the partial forces T h e second camp, of mainly electrical engineers, thinks experienced by the particle. (In matter the macroscopic that H is the primary magnetic field which causes the E and B fields are related to the wildly fluctuating microscopic fields by spatial averaging over an appropriate B field in magnetizable matter. macroscopic volume. Time averaging is not necessary. [la, This state of affairs, more than a century since Faradays pp. 226-2291.) field concepts were first presented in mathematical form by Maxwell, is surprising a t first. However, although some Lorentzs law is consistent with Einsteins special theory of the arguments for H as the fundamental field are su- of relativity since it can be derived from Coulombs law perficial there is also a deep theory formulated by Chu a t for the force between stationary electric charges, and EinhlIT in the fifties [2], which has been carefully supported steins two postulates [13, pp. 224-2311. It is classed as a by Haus and Penfield, e.g. [5, pp. xxii, 509-5101. classical law because it fails t o explain certain quantum mechanical phenomena, as demonstrated by the BohmIn t,his paper the arguments in support of the primacy Aharanov double slit experiment, and has t o be replaced of the B field are given first, a t an introductory level, by a more general force law in terms of the electric scalar by drawing in particular on the thinking of Feynman [l], and the magnetic vector potentials [l, pp. 11-15-8 to 141. Purcell [8] and Raab [9]. Then, in Section 6, Chus model in which the H field is considered to be the fundamental Clearly, in the International System of Units (SI), the elecmagnetic field is briefly outlined. tric Lorentz force term, f, = qE, defines the unit of the electric field, E, as being newton/coulomb (N/C). By inActually a t issue is whether the classical electromagnetic troducing the concept of electric potential and considering field should be represented by the pairing E - B or E-H. the potential energy of a charged particle in an electric (Ot,lier possibilities, such as the representation in terms field the unit for the electric field can also be shown to of die scalar electric potential and the magnetic vector be volt/meter (V/m). T h e rnagnet.ic Lorentz force term, potential [ l , p. 11-15-14]) will not be considered here.) f, = qv x B, defines the unit of the magnetic field, B, as newton/(coulomb meter/second). This physically meanDoes it really matter which point of view is held ? Certainly ! For scientific and philosophical reasons students ingful but cumbersome unit is by definition equal to the of electromagnetic theory, whether they be aspirant engi- tesla (T), the magnetic field unit which h a replaced the neers or physicists, deserve t o be taught which model is weber/meter2 (Wb/m2) in modern practice. The gauss most consistent with the laws of nature as understood a t (G) an early, non-SI, unit is convenient when working with present. If the two models are equivalent, or have com- weak magnetic fields. (The earths magnetic field is aptesla.) plementary strengths and weaknesses, then this must be proximately 0.5 gauss and 1 gauss = clarified. If one of the representations is significantly inferior then it should be rejected in our teaching and practice. The microscopic Lorentz force is manifest in a great variety of devices: in the ubiquitous cathode ray tube - J. J . Thomsons cathode ray being a unipolar current of electrons - the beam direction is controlled by electric and 2 The Lorentz force law defines magnetic fields; in magnetrons and travelling wave tubes; in the electron microscope; in the workings of diodes, tranfields the E and sistors and Hall effect magnetic field sensors; in particle accelerators such as the cyclotron; and, most notably, in T h e E and B fields are defined by the microscopic force moving free electrons through the vast global network of metallic conductors t o make energy available to humans equation f = qE+ qv x B (1) in the form of heat, light or mechanical work. which is associated with the name of Lorentz3. This turn In the latter context electrical engineers are familiar with of the century law of physics, which has been extensively the macroscopic form of the magnetic Lorentz force law, tested by experiment and practice, attributes the force f AF, = I x BAL, experienced in a n electromagnetic field by a particle with charge q and velocity v partially to the influence of the electric field E and partially t o the magnetic field B, [l, which is used to quantify the macroscopic force on a segment, of length AL, of a current carrying conductor [l, p. 11-13-1 and p. 11-15-14], [8, p. 2081. p. 11-13-31. It forms the basis for our understanding of Thus the Lorentz force law does two things [ l l , pp. 71- electrical machines and generators and is derived from the 741. It provides the link between classical mechanics and fundamental microscopic law.

Nobel physics prize, 1965: quantum electrodynamics. 2Nobel physics prize, 1952: nuclear magnetic resonance. Rindler [lo] provides interesting historical background on the Lorentz force law by tracing its magnetic component back to Heaviside.

355
0-7803-3019-6/96/$3.00 0 1996 IEEE.

Gauss?s law in dielectrics and the D field

For linear, isotropic matter the reiatioiisliip I>etween P and E may be written in the forin P = toxeE

Most, if not all. modern textbooks on electromagnetism for undergraduate courses in physics. and electrica.1 and electronic engineering, agree t h a t E is tslie primary electrical field while D is an induced field which is used t,o describe the interaction between the primary field, E, aiid electrically polarizable matt,er. The int,eraction resdts in electrical dipoles and a.lso the higher niultipoles (electric quadrupoles, electric octopoles etc.) being induced in the medium. In a general model t,lie contribut#ionof the higher order multipole moment,s [8. pp. 352-3551, [9], [la, pp. . . . . 136-146, 226-2351 to the polarization density must also be taken into account, but a discussion of this is deferred until Section 5. Thus the following development is based on the electric dipole approximation in which. from a ina.croscopic viewpoint, the interaction is modelled as a nett dei1sit.y of elect,rical dipole moments, P , which is induced in matt,er by t,he primary field.

-ye. This equation also makes it explicitly clear that

by introducing the macroscopic elect,rical susceptibility, E is t,he cause of P . and hence D . (The elect,rical susceptibility, x e , is related to the microscopic elect,rical polarizability of the constituent atoms or niolecules t8hroughthe (IlausiusMossoti, Lorentz-Lorenz and other niodels for the local electric field [ l , p . 11-11-7. 11-32-71, [12, pp. 152-1551. [13. 11. 3431, [14! p. Er].) Thus for linear5. isotropic dielectrics D can be written. in the electric dipole approximation, as

D = EOE P == t o ( 1 + x e ) E = tE = cot,.E. +
where the electrical I>eriiiit,t,ivityis defined in ternis of t,he electrical susceptibilit>-as E = tOcr = C O ( 1 ,ye) of the medium.

The constitutive relation D = c,JE,.E mliicli is coniiiioiily used in engineering practice unfortunat,ely olmxres t,he By distinguishing explicitly between t,he free and the fundament,al phxsical role of the polarization density P . bound charge densities, t.he lat.ter associa.ted with polar- Thus nmny student,s and engineers may not fully appreciate that D is a composite field iiivolving the primary ization, Gausss law can be written in the form field, E, and the induced polarizat,ion density, P . This is a pity since. besides being responsible for increa.sed capacitance as originally explained by Fa.raday, the polarization mechanism is exploited in a variety of important scientifThen the secondary field D , is defined by the linear super- ic and engineering applicat ious such as electrophoresis a.ncl posit,ion equat,ion the industrial scale electrical separators which break down oil-water emulsions and countless other mixtures [15]. The physical reality of the forces result,ing from the polarization mechanism have been clearly demonstrat,ed by Melcher [16, pp. 127-1301, [5, pp. -294-495, 514-5161. Nevertheand it satisfies a modified form of Gausss law., less, despite hiding the role of the polarization charge density, there is no implicit physical misrepresent,at,ion of t,he macroscopic electric polarizability phenomenon or mechacliv D = div ( E O E P ) = pfree, nisin in the use of D = cE provided the tlielect,ric may be taken to be linear and isotropic. explicitly involving only the free charge density. In the electric dipole approximation the polarization charge den- Authors like Feyiiinan [ l , pp. 11-10-7, 11-32-4 to 51 (see quote on the first page), Purcell [8, pp. 381-352, 4331, sity appears implicitly t,hrougli the relat.ion Shadowitz [6, pp. 269-2721 and Griffiths (4, p. 2601 do not. consider D to be very useful physical concept,. For esample, in practical electrocluasist,atics the electrical potential difference between the electrodes of capacitive systems is A variety of names for D abound in the literature. Ex- usually the independent varkble; and E is directly conamples are the electric flux density, the free electric trollable by the potential difference. T h e resulting distriflux density, the electric displacement, the electric bution of free charge, which determines div D , is usually displacement flus density. These terms, inspired by incidental and not a design parameter. Maxwells d i ~ p l a c e m e n t ~ current [G, p. 41 11, frequently cause more confusion than enlightenment , aiid the prac- However the result div D = 0 when pfree = 0, with the tice of Shadowitz [G, p. 2693 and Purcell [S, pp. 432-4331 resulting integral form fA D . da = 0, does provide a helpin simply referring to the D field may be preferable. A The electric susceptibility and permittivity concepts are ofcommon name for the E field is the electric field inten- ten implicitly extended into the non-linear, and possibly hysteretic sit,y, but liere the term electric field or just E field regimes, of dielectrics such as the ferroelect.ricmaterials but then it is impossible to talk of the susceptibility because i t is dependent on will be used in accordance with modern practice.

*See Purcells footnote regarding Maxwell being led astray by his displacement concept [8, p. 3821.

the amplitude of the electric field. Similarly for the magnetic susceptibility and the permeability of magnetic materials discussed in Section 4.

356
0-7803-3019-6/96~S3.00 0 1996 IEEE.

ful intermediate step when solving problems involving dielectrics where the free charge density in the material is negligible; especially so if the probleiii is highly syninietric. Of particuhr utility in boundary value problems is the coiitiiiuity of the normal component of D a t the interface between two different dielect,ric media i n the absence of free surface cha.rge. Finally, it is important to note that D = EOE,E a is frequency domain representation of the dielectric properties of a material. It is implicit, that the medium is being excited time harmonically at. a certain frequency and in this representation the permittivity E, ( U ) is complex to account for dissipation, arid frequency dependent because all physical materials are dispersive. There is a t present great! interest in the behaviour of materials subject, to impulsive excitation and a variety of phenomena are being studied in the tinie domain using experimental, analytical aiid numerical methods. Time domain analysis followed by Fourier transformation of the results has also become a popular method of obta.ining frequency domain data. The representation of D as the product of E with the frequency domain permitt,ivity leads to a cumbersome time domain convolution integral representation in which the permittivit,y impulse response of the material is the Fourier transform of cT(w). N o such coniplicat.ions arise in transforming the funda.menta1 form D = toE P to the time domain after invoking Lorentz or Debye models. In particular this representation now seems t,o have significant adva.iitages in the numerical modelling of dispersive ma.t#erialsby means of finite difference t,ime-doniaiu algorithms [17], [18].

After introducing the notion of inagnetlic susceptibility and permeabilit,y, the appropriat,e form of the magnettic cond t u t i v e relatioil will be shown to be not B = p H but H = pL-lB. Ampkres law in niatt.er can be stated in the differential form curl B - p o ~ o d E / B t= pO.Jtotal xhicli involves the B a i d E fields a t a macroscopic ..point. the vacuum parameters E O and po, arid Jtotal, the total current density at t,liat point. In matter which contains free charge carriers, and which has dielect,ric and magnetic properties. the total current density may be decomposed and expressed as t.he sum of the free current density. Jfree, and t.he bound current density, J b o u n d . In turrl t>he bound current densit.y can be considered to be the suni the polarization current density, Jpo19 and the magnetization, or -linpi-rian, current density, Jmag. Thus
Jtotal

= Jfree

+ Jbound

Jfree

f Jpol

+ Jmag.

(5)

An engineering example of a material in which all three t.ypes of current. densities might flow is a microwave absorbing comp0sit.e consisting of a fine magnetite powder embedded in a polyurethane host with die magnetrite particles packed close enough to make contact with each other arid thus allow ohmic currents to perco1at.e.

So Ampi-res law i n matter can be written in the general form

Amp&reslaw in magnetizable matter and t h e H field

By pursuing the development in Sectioii 3 the polarization current density, in t,he electric dipole approximation, can be relat,ed t.o the electric polarizatiori density P by

Whereas clarity exists rega.rding the relationship bet.ween E and D there are, as mentioned in the Introduction, two schools of thought regarding the relationship between the B a i d H fields. One6 superficial argument is based on the a.pparent analogy between the popular constitutive relations D = tE and B = pH which seem to pair E with H; hence the conclusions that H is t,he fundamental magnetic field, and that the magnetic permeability p is analogous to the electrical permittivity E. In this Section it is taken as a basic postulate that the Lorentz force law defines B as the fundamental ma.gnetic field. It is also assumed7 t h a t only magnetic dipoles contribute to the magnetization density M. The general relationship between B and H is then developed from Ampkres law [ l , pp. 11-36-1 to 61, [4, pp. 251-264, 304-3111, [G, pp. 306-3211, [8, pp. 423-4371 and [13, pp. 408-4121.
Another is that the ratio E I H has the unit of impedance, QED. The approximation in Section 3 that P is only due to electric dipoles, and which resulted in the definition D = coE+P, seems to be consistent with the magnetic dipole assumption of this Section. However, as discussed in Section 5 , this is not the case because electric quadrupoles contribute to the same order as magnetic dipoles
[9, 191.
.-

It can also be shown that, in the magnetic dipole approxima.tion, the magnetization current density is relat,ed to t,he magnetization density, M. by
Jmag curl M =

(8)

Thus Ampkres law can be rearranged into the form curl ( p ; l ~ - M) - ~ ( E , , E


Now the H field can be defined as

= J~~~~

in terms of the B field and the iiiagnet,ization density, M. Also, the term EOE P on t,he right hand side may be replaced with the D vector which was defined in Sectmion 3. This yields Ampkres law in terms of H and D as given in the Introduction

0-7803-3019-6/96/$3.00 8 1996 IEEE.

3 57

From the viewpoint developed in this Section the status of - M is no different from t h a t of D = EOE P . Both are considered to be auxiliary, or induced, fields which conveniently summarize, but also obscure, the interaction between the fundamental fields, B and E, and matter .

H = p;lB

quivalent. This is not the case for ferromagnetic or ferrimagnetic materials. For them it is convenient t o use the conventional definition M = x m H because, as shown above, the H field under magnetoquasistatic conditions is completely determined by the free currents and is independent of the material properties, whereas the Ampkrian current density Jmag curl M , which is dependent on the = This view does not detract from curl H - d D / d t = Jfree material properties, also contributes to the B field. Then, being a powerful statement of Amptres law. This is par- after defining the relative permeability of the material ticularly so when dealing with engineering applications in- as p,. = 1+xm the magnetic material constitutive relation volving ferromagnetic and ferrimagnetic materials a t fre- is H = p;(l+ xm)-lB = po p,. -1 -1B-- p - B . quencies which are sufficiently low t o regard the system as magnetoquasistatic (MQS) in the sense that aD/at x 0 [5, pp. 70-711. T h e utility of the MQS approximation Much confusion as t o the status of B ,uis-a-vis H arises curl H x Jfree evident by considering the integral form because this relation is commonly presented as B = pH, is a form which Elliott [7] regards as being an historical accident, and because the magnetization properties of ferH . ds M Jfree . da = Ifree. romagnetic and other non-linear and hysteretic materials are often presented as B-H graphs a i d not as B - I f r e e It is now clear t h a t the line integral of H around any closed graphs. See Shadowitz [ G , pp. 312-3211 and Purcell [8, pp. path C ; whether that path passes through a =-linear 422-423, 435-4371 for additional perspectives on magnetic magnetic material like iron or not, is dependent solely on susceptibility and permeability. the free current enclosed by the path. T h e material properties are irrelevant. This fact is of great utility in dealing with the engineering applications of ferromagnetic and ferMaxwells equations and multirimagnetic materials, with their non-linear and hysteretic 5 properties, in the design of the magnetic circuits of elecpole moments tromagnets, transformers or machines, e.g. [l, pp. 11-36-6 to 111. Here the free currents and their paths are primary design variables. Purcell [8, p. 4331 and Griffiths [4, p. By using Equations (2) and (6) t o explicitly show the con2601 nicely discuss the utility of H in the context of the bributions of the bound charge and current densities the free current being under the control of the designer. They D and H fields can be eliminated from the laws of Gauss also use the opportunity, as also mentioned in Section 3, and Amptre and Gauss in Section 1, e.g. [4, pp. 307-3111. to juxtapose the D field as being of minor importance Thus because in electrostatic or electroquasistatic systems the free charge density is rarely a design variable. Rather, it curlE+dB/dt = 0 is the electric potential, and thus t h e electric field, which is directly controlled. curl B - poodE/dt = ,%(Jfree 4-J b o u n d )

S,

Nevertheless, despite its practical utility in solving magnetic circuit problems, H is the auxiliary magnetic field of the first camp. Thus their preferred nomenclature is that B is the magnetic field instead of the magnetic flux density, or magnetic induction of historical tradition. Also, for H just the name H field instead of the traditional magnetic field intensity [l, pp. 11-32-4 to 5, 11-36-12], [4, pp. 260-2611, [6, pp. 312-3211, [8, pp. 431-4371,

div EoE = Pfree div B = 0

Pbound

Then from Equation (4) it would seem that the bound charge density is Pbound = -div P , while Equations (5), (7) and (8) suggest t h a t the bound current density is Jbound = aP/& curl M. These conclusions are however flawed. Raab and his colleagues have proved [9] that if the contribution of magnetic dipoles are taken into account, as they are in the expression for J b o u n d by the term curl M , By analogy with the electric susceptibility xerSection 3, then the electric quadrupoles must also be accounted for. the magnetic susceptibility can be defined by relating the Thus, using tensor notation, the correct form for the imagnetization M t o the B field by the relation M = component ( i stands for z, y or z ) of the bound current p i l & B . By this definition H is expressed in terms of B density in the magnetic dipole-electric quadrupole approxas H = p i l ( l - x:,)B. However it is convention to define imation is [9, eqn. 341 the magnetic susceptibility as M = xmH, which yields the constitutive relation H = p;(l xm)-B. The P f i j k v j h l k - $vjQij-k i Jhound, two definitions of magnetic susceptibility are related by (1 - x:,~) (1 x,,~)- and since (1 x,~)- (1 - x m ) This is just J b o u n d = a P / a t + curl M with the contribu= w if << 1 it is evident t h a t for paramagnetic and dia- tion of the time varying electric quadrupole density, Qij magnetic materials the two definitions are essentially e- addedg. If this is not done the resulting hIaxwells equa-

See the interesting remark attributed to Sommerfeld concerning Maxwell and H [4, p. 260, footnote]

The dot notation is used to indicate partial derivatives with respect to time, and c l j k , is the alternating tensor which allows vector products to be written in tensor form.

0-7803-3019-6/96/1F3.000 1996 IEEE.

358

tions are not origin independent, in violation of a fundamental requirement of physics. (The laws of physics may not depend on an arbitrary choice for the origin of the reference frame.) The underlying reason for the violation of t*ranslational symmetry is t h a t the relative magnitudes of the multipole contributions to an electromagnetic effect are ordered as [SI, [ l a , pp. 391-401 and Chapter 161 electric quadrupole magnetic dipole

naturally to the electromagnetic field being represented by pairing E with H: not B . Chu postulates [2, Sections 5.4, 7.10-11 and Appendix 1 1 that the predictions of his magnetic-charge dipole model can not be distinguished (with particular emphasis on ferromagnetism. ferrimagnetism and paramagnetism) by external field or force measurements from the predictions of the Ampkrian current loop model. Thus: the argument goes, the magnetic-charge dipole model is preferable because of its simplicity (no moving parts in the rest frame of the material [23]) and because it allows the treatment of magnetic polarization phenomena to be developed by direct analogy with electric polarization. Chu also takes issue specifically with the Ampilrian current density Jmag = curl M, Equation ( 8 ) , arguing that its use in Poyntings theorem leads to inconsistencies. Tellegen 1241 disagreed, giving reasons, and in turn he raised an objection to Chus model based on an apparent difference between the predicted forces on a magneticcharge dipole and a microscopic Ampbrian current loop. However, Haus and Penfield [25], [5, pp. n i i , 509-5101, [26, Section VI11 have shown that Tellegen had neglected a relativistic effect and that the two models in fact predict the same force for a magnetic dipole of constant moment in. And in t,lieir 1967 book [27] they have apparently shown [23] that the same magnetic force densities are predicted by choosing either B or H as the fundamental magnetic field. However Shadowitz [6, p. 3151 still demurs.

electric dipole

>>

>>

Thus electric quadrupoles contribute effects to the same order as magnetic dipoles. The correct forms for the D and H fields t o this order are [9, eqns. 37 and 381

Here the expression for H corresponds exactly to Equation (9) but the expression for D differs from Equation (3) because the contribution of the electric quadrupole density was neglected in the latter. Raab and Cloete [20] have shown, for non-ferromagnetic matter, that when modelling the electromagnetic behaviour of anisotropic chiral matter the electric quadrupole contribution to the predicted optical activity is approximately the same as that of the In his 1989 paper Rindler [lo], a,pparently unaware of the magnetic dipoles. This has been confirmed, both theoret- work of Chu and the ensuing controversy, again raises the ically and experimentally by Theron and Cloete [21, 221. problem of the equivalence of the magnetic-charge and curThus care must be taken to correctly include the correct rent loop dipoles. He concludes that the magnetic-charge multipole contributions in J b o u n d and Pbound, with the or- pair dipole gives exact answers for the torque on small der of the multipole expansion depending on the physical current loops a.nd, under certain circumstances, also for the corresponding force. However it is not clear from phenomena which are under investigation [9]. a superficial reading whether his analysis has the depth Complete expressions for the multipole moment densities and generality of the Haus and Penfield treatment of the Pi,Qlj and llfi in terms of the E and B fields and their problem [as]. Thus the current status of the literature on spatial and time derivatives are given in [19, eqns. 9-11] Chus model seems to be that nobody has formally proved to the order of electric quadrupoles and magnetic dipoles. it to be physically inconsistent within the espistemological These can then be used with the above expressions for D, ambit of classical electromagnetic theory. and H;to obtain physically rigorous constitutive relations.

Chus E-H model

7 Conclusion. B or H ?
Some of the reasons for considering B t o be the fundamental magnetic field are given by Raab [28].

The 1960 text book by Fano, Chu and Adler [2] was written for use by third year students of electrical engineering a t MIT. T h e material is based on an E - H model of the electromagnetic field which was developed by Chu and which is presented in the book. In their 1989 textlo, also for use by undergraduate students of electrical engineering at MIT, Haus and Melcher [5] adopt Chus formulation of electrodynamics which uses a magnetic-charge dipole model, not the generally accepted microscopic current loop or Ampkrian model, for magnetization. This approach leads
It. is dedicated t o Adler, Chu and Fano

1. B (and E ) as defined by the Lorentz force law, Equation l , is unique and origin independent in vacuum and also in matter, the latter by spatial averaging over a macroscopic volume. (Although the posit,ion, I*,of the charge is origin dependent its velocity, v , and acceleration are not because the time derivative of an arbitrary but constant offset is zero. Acceleration is mentioned because it is proportional to the force in
I have been unable to obtain a copy of the book.

0-7803-3019-6196/$3.00 0 1996 IEEE.

3 59

est,ernal field measurement . . . (or) by external force nieasurement,s. Also, according to Haus [23] the magneticcharge dipole is a inucli simpler model because it contains no moving part,s . However the Einstein-de H a m esper2 . H enters the E-B formuhtion through Ainpkres law, ilnelit does nlov.ng W I ~ magllet,izatioll ~ ~ ~ Equation 1 and it is not uniquely defined in matter of a bar of iron, suspellded 0 and coaxially ill a because its divergence is arbitrary. solenoid, is reversed by changing the direction of curren0 3. By using the appropriate niultipole expansions, as dis- t, flow through t,he solenoids windings the bar be, 111s to rot,ate about its axis if the experiment is very carefully cussed in Sect,ion 5, Amplres law can be written in a origin independent form. However H (and D) is itself in general not origin independent as can be seen from Einstein interpreted this inacroscopic effect, tlie gyromagit,s expansion in tmerlns multipole moments - the moof by lvriting .\$re have given firlll proof of inelits depend 011 the choice of origin. (Graham and of ~ l l l p ; r e ~ lllolecu~arCUrrelltS [30, PD. 245s Raab lave recelltlY derived an Origin indePeli- 2.16], The 1nodern illterpretat,ioli of tlie EinsteiIl-de Haas dent form: in the 11lagllet,icdipole-elect,ric WadruPole effect is in t,ernls of electroll spin and associated to describe lie boundary approximatiolll for momentum, a much more abstract quantum mechanical Alllpkre.s electric [30. pp, ditions at the interface of b e h e e n the vacuum and a concept n o n - n q n e t i c chiral mediuiii. However this formula- 246-2471. N ~ ~ , ~ ~ ~i ~ ~~ ~ ~ effect, does ~~ ~ ~~ Ii llaas ~ generally aI.I)licable to forms Of tion is seelll to veto t,lle Cllus lllaglletic-charge dipole Jf,it,]i llo non-relativistic (Newtonian) mechanics.) Origin independence of its laws is a fundamental requirement of physics, as mentioned earlier.

~ ~

~ - d

E 4. Tile special TI^^^^^ of relativity s~lOvvs B are illtilnately alld ~ulldalllenta~~y lillkec1, e,g, [I, Sections 11-13-6 alld 7. cllapters and 261, [13, 11-25 Chapters 4 and 51.
pp. 1085-10861 for fur5 . See [:32, pp. 21-27] and [20, ther points concerning ;\laswells equations, Lorentz covariance and the constitutive relations.

moving part.s, and thus presuinably 110 intrinsic a.iigular momentum. This would in turn significa.ntly undermine t8hephysical founda.t,ion of Chus E-H formulation for the classical electromagnet.ic field.

Acknowledgements

I wish to thank Professor R,oger Raab for ma.ny discussioiis on tlie topic of t,liis paper. in particular for introducing me t.0 the multipole representation of the interacttion heH ~ cllus ~H field call Ilot he crit,icized oll the t,ween the electromagnetic field and niatt.er, and for always ~ ~ ~ ~ , grounds of origin dependence or llon-uniqueness because finding time t o reveal some of his deep pl:ysica,l insight to his definition [2, p. 121, [5, pp. 364-3651 of H is in terms me. of the Lorentz force law, albeit. t,he following non-standard I also wish to thank Professor Hermann Haus for his illuform, minating communicat.ion [23].
Thus the case for B as the fundamental magnetic field in t,he st,andard E-B formulation is ext,remely strong.

f =qE+qv xpoH,

and not via Ampkres law. Here is tlie vacuum con- Xly colleague hlr. KOOS Holtzhausen provided the pun stant and so poH in the Chu formulation is identical to from Hamlet ! B in the standard formulat,ion. In his formulation the B field is defined, via Faradays law, as B = po(H M) and it, is interpreted to be a secondary field needed only References for dealing with magnetizable matter. This definition is directly analogous to the definition of the D field i n terms [1] R. P. Feyninan t-t a1 , The Feynnian Lectures on of the electric dipole density as D = toE P . Although Physics. Redwood City: Addison-\Yesley, 1964. Chu and Haus were apparently not aware that this pairing [2] R. M. Fano, L. J . Chu and R,. B. Adler, Electronzagby analogy of electric and magnetic dipoles is flawed, as discussed in Section 5, the definition of D can be rectiuelac Fields, Energy, w i d Forces, New York: John fied by including the electric quadrupole coutributioiis to Wiley and Sons, 19G0. make the definitions consistent in the magnetic dipole and [3] S. Ramo, J . R. \.lhinnery and T. van Duzer, Fields electric quadrupole approximation. and Waves an Coniniunacaiaoii Electronacs, Third Edition. New York: John Wiley and Sons, 1993. Thus the question arises whether there is any experiment which vetos Chus magnetic-charge dipole model in favour of ~ ~ current ~loop nlodel, ~l~~ ~ i i ~ ~ ~ D. J . ~ i ~ - d ~ ; ~ [4] ~ ~ Griffiths, Iniroductzon t o Electrodynamzcs, Sec~ ond Edition, Englewood Cliffs: Prentice Hall, 1989. Haas experiment of 1915 seeins t,o do so 130, pp. 245.. 2491, [31, PP. 449-4551? [I, P. 11-37-51. As 111entioned in [5] H. A. Haus and J . R. hilelclier, Electronzagnefzc Fzelds Section 6, Chu selected the magnetic-charge dipole model a n d Energy, Englewood Cliffs: Prentice-Hall, 1989. instead of the A m p h i a n current loop model partly because [a, . 1751 the two models cannot be distinguished by any p

0-7803-3019-6/96/$3.00 1996 IEEE. 0

3 60

[21] I.P. Tlieron and J.H. Cloete. The optical act,ivity of an artificial non-magnetic uniaxial chiral crystal a t microwave frequencies, Journal of Electromagnetic Waves and Xpplica.t,ions, vol. 10, no. 4 , pp. 539-561, [7]R. S.Elliott, Some useful analogies in the t,eacliiiig ~ ~ 1996 ~ i 1 of electroniagiiet,ic theory, IEEE Trans. Educat.ion, vol. E-22. no. 1. pp. 7-10, February 19i9. (Sce also [22] I.P. Tlieron and J . H . Cloete. The electric quadrupole contribution to the circular birefringence of nonmagR. S. Elliott, Electromagnetic theory: a siiiiplified netic a.nisot,ropic cliiral media.: a circular waveguide representation, IEEE Trans. Education. vol. E-24. esperimeiit, IEEE Transactions on Microwave Theno. 1, p p . 294-298, November 1981.) ory and Techniques. Scheduled for publication in vol. [8]E. hI. Purcell, ElectricLty u n d illngaetism, Second 44, no. 8.August 1996. Edit,ioii. New York: RIcGraw-Hill, 1985. [23] H . A. Ha.us. H versus B, Private communication by [9] E. R. Crraliarri! J . Pierrus, arid R. E. Raab, hlultielectronic nia.il, 12 J u l y 1996. pole moments and Rlaswells equations. J Physics [24] B. D. 1. Tellegen, hla,gnetic-dipole models, AmerB , vol. 25, pp. 4673-4684. 1992. icaii Journal of Physics, vol. 30, no. 9 , pp. 650-652, [lo] IV. Rindler, Relativit,y and electroniagiietisiii: The Sept,ember 1963. force on a magnetic monopole. A m . .J. Pliys, vol. :j7. [25] H . A . Haus and P.Penfield, Force on a conducting no. 11, pp. 993-994, November 1989. current loop, Physics., vol. 42, pp. 447-4511, 1969. [ l l ] C:. IT. Misner, I<. S. Thorne and J . A. Wheeler. Grnvitntzon. New York: \V. H. Freemaii and Company, [26] H. A . Haus and J . R. hlelclier, Fields that are always 1973. dyna.mic, IEEE Tra.ns. Education, vol. 3 3 , no. 1 pp. 35-46. Februa.ry lC)90. [12]J . D. Jackson, Classical Electrodyriamics, Second EdiSee also H. A. Haus and J . R.Melcher: Electric ancl tion, New York: John IViley arid Sons, 1975. iiiagnetic fielcls, Proc. IEEE, vol. 59, no. 6 , pp. 887894, June 1971. [13] R. S. Elliott, Eleclrolrrtzyiieiics: History, Theoryl and

[ii] A . Sliadowitz, The Electromagnetic Field.New York. AicGraw-Hill, 1975. (Reprinted by Dover Publications in 1988.)

[27] P. Peiifield and H. A. Haus, E l ~ t r c ; d y ? i . u i ~ofcMOVi ~ ing d l e d i n . Cambridge: MIT Press, 1967 [14] M.Born and E. TVolf. Principles of Optics. Perganion Press. London. 1959. [ 2 S ] R. E. Ra.ab, Digression on the 13-13 cont,roversy, P r ivat8econiniuni ca t io 11. 1-10c t o b er 1994. [15] L.C. Waterman, Electrical coalescers. Chemical Engilleering Progress- vol. 61, 110. 10, PP. 5l-57, oc- [29] E, B, Grallalll and R, E.Ra,ab, Reflection from 11ontober 1965. centrosymmetric uniaxial crystals: A multipole a.pproach, .J Opticti1 Soczely of Ainericu A , accepted [I61 &I. Zahn and H . A. Haus, Contribut,ions of Prof. for publication October 1995. (Additional det,ails are James R. r\lelcher to engineering education, Jouriial available in .New insights into the Ma.xwel1 bounclary of Electrostatics. vol. 34. pp. 109-162, 1995. conditions, private communication, 10 March 1995.) [17] J . L. Young, Propaga.tion in linear dispersive media: finite differelice t~me~~ollla,ill nlet~lodologies,~~ IEEE [30] A . Pais, Subtle 2s fhe Lord - t h e science a n d life ofAl6e7.i E27rslei72, Oxford: Oxford TJniversity Press, 1982. Trans. Antennas and Propagation, vol. 43, no. 4,pp. 422-426, April 1995. [31] R. M. Rozroth. F~rroinngiiet,ism,New York: IEEE Press, 1993. [18] J . B. .Judliins and R. W . Ziolkowski, -Finite-difference
hne-tluriiaiii modeling of iioiiperfectly conducting metallic thin-film gratings, Journal Optical Society of America A , vol. 12, no. 9, pp. 1974-1983, Septeniber 1995.
[19] E. B. Graliam and R. E. Raab, Magnetic effects in

Applicatzons, New York: IEEE Press. 1993.

1321 T. H. oDell, TiLe e~ectrodyllanzicsof magnebo-electrtc media, Arris t er darn: North- Holland, 19 70.

antiferromagnetic crysta.ls in the electric quadrupolemagnetic dipole approxiiiia.tioii, Philosophical Magazine B, vol. 66, no. 2. pp. 269-284, 1992.

[20] R.E. Raab and J.H. Cloete, An eigenvalue theory of circular birefringence and dichroism in a nonmagnetic chiral medium, Journal of Electromagnetic Waves and Applications, vol. 8, no. 8, pp. 1073-1089, August 1994

0-7803-3019-6/96/S3.000 1996 IEEE-

361

S-ar putea să vă placă și