You are on page 1of 7

Electromagnetic Fields of a Rotating Shell of Charge

Kirk T. McDonald Joseph Henry Laboratories, Princeton University, Princeton, NJ 08544 (April 3, 2003; updated August 10, 2008)


A spherical shell of radius a carries charge Q uniformly distributed over its surface. Give expressions for the electric eld E and magnetic eld B everywhere in space when the shell it rotating with angular velocity about an axis. Compare these elds with those reported by an observer in a frame that rotates with angular velocity about the same axis. c for any relevant It suces to consider the low-velocity case, where r c and r radius r.


This problem will illustrate a remark by Schi [1] that care must be taken when discussing electrodynamics in rotating frames, because the eect of the distant stars is to introduce ctitious charges and currents into the problem. The example chosen here for a spherical shell of charge is perhaps slightly more complicated algebraically that the case of a cylindrical shell of charge, but the latter is less consistent with a universe in which distant stars play a role.


Rotating Spherical Shell in the Lab Frame

We work in a spherical coordinate system (r, , ) with origin at the center of the sphere, and z axis along the axis of rotation, such that the angular velocity of the spherical shell is . Then, in Gaussian units, the electric eld is simply = z E(r < a) = 0, Q r, E (r > a ) = 2 r


independent of the angular velocity . If that angular velocity is nonzero, there is also a nonzero magnetic eld in the lab frame. A calculation of the magnetic eld via the vector potential is given in sec. 5.4.1 of [2]. For possible contrast, we utilize a scalar potential that can be dened everywhere except on the shell of radius a, since B = 0 every where except on that surface. Thus, we write B = , (2)

where the potential is not continuous across the surface r = a because of the surface currents there. The potential is azimuthally symmetric, should be nite at the origin and

should vanish at innity. Suitable forms for the regions inside and outside of the sphere are therefore (r < a) = (r > a) = rn Pn (cos ), an an+1 Bn n+1 Pn (cos ). r An


We can obtain two matching condition across the surface r = a from Maxwells equations. First, B = 0 implies that the radial component of B, ie, Br = /r, is continuous at r = a. Hence, nAn (n + 1)Bn Pn (cos ) = Pn (cos ), (4) a a and therefore n+1 An = Bn . (5) n Also, Amperes law, B = 4 J/c, applied to a small loop in a plane of constant that surrounds a segment of the charged surface, tells us that 1 (r = a+ ) 1 (r = a ) 4 K = B (r = a+ ) B (r = a ) = + , c a a where the surface current density K due to the rotating charge is given by K= Q 1 Q sin Q d cos Q dP1 (cos ) 2 2 a sin d = = = . (7) 4a2 2 a d 4a 4a d 4a d Q P1 = (r = a+ ) (r = a ) = c (6)

Using eq. (7) in (6), we can integrate the latter to nd,1 (Bn An )Pn . (8)

Since the Pn are a set of orthogonal functions, we see that An = Bn = 0 for n = 1, and that in view of eq. (5), Q = B1 A1 = 3B1 . (9) c Thus, 2Q Q A1 = , B1 = , (10) 3c 3c and 2Qz 2Q rP1 = , 3ac 3ac Qa2 cos (r > a) = . 3c r2 (r < a) = It is useful to note that the magnetic moment m of the spinning shell of charge is z m= c


Q z Qa2 2 2 ( a sin ) 2 a sin d = 4a2 2 4c

Qa2 z Qa2 = . (12) sin d = 3c 3c


Thanks to Marco Moriconi for pointing this out.

The potential can now be written 2m z, a3 m cos (r > a) = . r2 (r < a) =


Thus, inside the shell the magnetic eld is uniform, and outside the shell it has the form of a point dipole whose strength is the total magnetic dipole moment of the rotating shell,2 B (r < a ) = 2m , a3 3(m r) rm . B (r > a ) = 3 r



Electrodynamics in a Rotating Frame

Summary of the Principles

We desire the elds E and B seen by an observer in a frame that rotates with angular is parallel to the angular velocity velocity with respect to the lab frame, where = z of the spherical shell in the lab frame.3 We restrict our attention to points r in the rotating frame such that the velocity v = r of the points with respect to the lab frame is small compared to the speed of light. Then, we can largely ignore issues of whether rods and clocks in the rotating frame measure the same lengths and time intervals as would similar rods and clocks in the lab frame, and whether the speed of light is the same in both frames. That is, we ignore Ehrerfests paradox [7]. A discussion by the author of electrodynamics in a slowly rotating frame appears in the lengthy Appendix to [8]. For reference, we reproduce the principles of electrodynamics in the frame of a slowly rotating medium where and dier from unity.4 The (cylindrical) coordinate transformation is r = r,

= t,

z = z,

t = t,


Another derivation is to note that the surface current (7) is the same as would hold for a uniformly magnetized sphere of radius a and magnetization M = Q/4ac. Then, a solution based on a scalar potential for the eld H rather than B [3] quickly leads to the result that B(r < a) = 8 M/3 = 2m, where the total dipole moment m = 4a3 M/3 = Qa2 /3c is the same as that calculated in (12). Again, the magnetic eld outside the sphere is that due to a point dipole of strength m. 3 In many examples of electrodynamics of rotating systems we are less interested in the elds observed in the rotating frame than those in the lab frame. Yet, it may be that knowledge of some aspects of the elds in the moving frame is helpful in reaching an understanding of the elds in the lab frame. In this case it is often convenient to characterize the elds in the moving frame via the use of Lorentz transformations between the lab frame and an inertial (nonrotating) frame in which some point of the rotating system is at rest. For reviews of this approach, see [4, 5]. A pedagogic example is given in [6]. 4 This case is discussed most thoroughly by Ridgely [9, 10], but primarily for the interesting limit of steady charge and current distributions.

where quantities in observed in the rotating frame are labeled with a . The transformations of charge and current density are = , J = J v, (16)

where v (v c) is the velocity with respect to the lab frame of the observer in the rotating frame. The transformations of the electromagnetic elds are v v B = B, D = D + H, (17) E = E + B, H = H. c c The transformations of the electric and magnetic polarizations are v P = P M, (18) M = M, c if we regard these polarizations as dened by D = E + 4 P and B = H + 4 M . The lab-frame bound charge and current densities bound = P and Jbound = P/t + c M transform to bound = P Jbound 2 M v + M, c c P P v M = + c M + v ( P ) + + P . t c t (19) (20)

Force F is invariant under the transformation (15). In particular, a charge q with velocity vq in the lab frame experiences a Lorentz force in the rotating frame given by F =q E + vq B c =q E+ vq B = F, c (21)

where vq = vq v. Similarly, the Lorentz force density f on charge and current densities in the rotating frame is J J + Jbound B = (free + bound) E + free B. c c Maxwells equations in the rotating frame can be written f =E + B = 0, D = 4free,total = 4 (free + other) , E + (22)

(23) (24)

B = 0, (25) ct D 4 4 = (26) H Jfree,total = (Jfree + Jother) , ct c c where free = free and Jfree = Jfree freev are the free charge and current densities, and the other charge and current densities that appear to an observer in the rotating frame are other = Jother v D v Jfree H + , 2 c 2c 4c ct D H D v = freev + . 4 4 4c t 4 (27) (28)

The other charge and current distributions are sometimes called ctitious [1], but we nd this term ambiguous. For an example with an other charge density H /2c in the rotating frame, see [11]. Maxwells equations can also be expressed only in terms of the elds E and B and charge and current densities associated with free charges as well as with electric and magnetic polarization: (29) E = 4total, and B where total = free = = more = Jtotal = = = Jmore = v D v H Jfree P + 2 c 2c 4c ct free,total P free + bound + more , v P B v E 2 Jfree + + c M + , c t 2c 4c ct P D v D H Jfree + + c M + freev + t 4 4 4c t P Jfree,total + + c M t Jfree + Jbound + Jmore , 2 M v v free P + M c c E v E B + . 4 4 4c t E 4 J , = ct c total (30)

(31) (32)



The contribution of the polarization densities to the source terms in Maxwells equations in much more complex in the rotating frame than in the lab frame.Because of the other source terms that depend on the elds in the rotating frame, Maxwells equations cannot be solved directly in this frame. Rather, an iterative approach is required in general. The constitutive equations for linear isotropic media at rest in the rotating frame are D = E, in the rotating frame, and D = E + ( 1) v H, c B = H ( 1) v E, c (36) B = H ( 1) v E, c (35)

in the lab frame. The lab-frame constitutive equations (36) are the same as for a nonrotating medium that moves with constant velocity v with respect to the lab frame.

We can also write the constitutive equations (35) for a linear isotropic medium in terms of the elds B , E , P and M by noting that D = E + 4 P and H = B 4 M , so that P M = = 1 E, 4 1 B 1 4

v E 1 = 1 c 4

1 v B P. 4 ( 1) c


Similarly, the constitutive equations (36) in the lab frame can be written to order v/c as P = M = 1 v 1 E+ 4 c 1 B 1 1 4 B 1 1v = E+ M, 4 4 1 c 1 B E 1 v v = 1 P. c 4 4 ( 1) c


Ohms law for the conduction current JC has the same form for a medium with velocity u relative to the rotating frame as it does for a medium with velocity u relative to the lab frame, u u (39) JC = E + B = E + B = JC , c c where is the electric conductivity of a medium at rest. 2.2.2 Fields of a Rotating Shell of Charge in a Rotating Frame

While the source charge and current densities and J are readily transformed from the lab frame to the rotating frame using eq. (16), it is not straightforward to solve Maxwells equations in the rotating frame because the other charge and current densities in eqs. (31)(34) depend on the as-yet-unknown elds in that frame. However, in the present example, we can simply transform the elds (1) and (14) to the rotating frame using eq. (17), with the results E (r < a ) = E (r < a ) + v v 2m B (r < a ) = 3 , c c a Q r) rm v v 3(m + , E (r > a ) = E (r > a ) + B (r > a ) = 2 r 3 c r c r 2m , B (r < a ) = B (r < a ) = a3 3(m r) rm . B (r > a ) = B (r > a ) = 3 r



where v = r. Even when = so that the electric charge appears at rest in the rotating frame, a nonzero magnetic eld (41) is observed in that frame, which can be attributed to the ctitious currents.

[1] L.I. Schi, A Question in General Relativity, Proc. Nat. Acad. Sci. 25, 391 (1939),

[2] D.J. Griths, Introduction to Electrodynamics, 3rd ed. (Prentice Hall, Upper Saddle River, NJ, 1999). [3] J.D. Jackson, Classical Electrodynamics, 2nd ed. (Wiley, New York, 1975), sec. 5.10. [4] J. van Bladel, Relativistic Theory of Rotating Disks, Proc. IEEE 61, 260 (1973),

[5] T. Shiozawa, Phenomenological and Electron-Theoretical Study of the Electrodynamics of Rotating Systems, Proc. IEEE 61, 1694 (1973),

[6] K.T. McDonald, The Wilson-Wilson Experiment (July 30, 2008),

[7] . Grn, Relativistic description of a rotating disk, Am. J. Phys. 43, 869 (1973),

[8] K.T. McDonald Electrodynamics of Rotating Systems (Aug. 6, 2008),

[9] C.T. Ridgely, Applying relativistic electrodynamics to a rotating material medium, Am. J. Phys. 66, 114 (1998),

[10] C.T. Ridgely, Applying covariant versus contravariant electromagnetic tensors to rotating media, Am. J. Phys. 67, 414 (1999),

[11] K.T. McDonald, The Barnett Experiment with a Rotating Solenoid Magnet (Apr. 6, 2003),