Energy and linear and angular momenta in simple electromagnetic systems

Masud Mansuripur
College of Optical Sciences, The University of Arizona, Tucson, Arizona 85721
[Published in Optical Trapping and Optical Micromanipulation XII, Proceedings of SPIE 9548, 95480K~1-24 (2015)]

Abstract. We present examples of simple electromagnetic systems in which energy, linear momentum,
and angular momentum exhibit interesting behavior. The systems are sufficiently simple to allow exact
solutions of Maxwell’s equations in conjunction with the electrodynamic laws of force, torque, energy,
and momentum. In all the cases examined, conservation of energy and momentum is confirmed.

1. Introduction. This paper presents several examples of simple electromagnetic (EM) systems,
whose complete understanding requires careful attention to the detailed behavior of EM force,
torque, energy and linear as well as angular momentum. The concept of “hidden momentum”
appears in some of the examples, which helps to illustrate the differences between the Einstein-
Laub formulation and the standard (i.e., Lorentz) formulation of classical electrodynamics [1-3].
In Sec.2 we analyze the energy and angular momentum content of a finite-duration, finite-
diameter wave-packet (or “light bullet”) propagating in free space and carrying both spin and
orbital angular momentum. Among other findings, we obtain a simple formula for the total
angular momentum per photon of the light bullet. The spin contribution to the total angular
momentum per photon turns out to be ℏ𝝈𝝈𝑧𝑧 , where 𝝈𝝈𝑧𝑧 = 𝒌𝒌𝑧𝑧 ⁄𝑘𝑘0 = 𝑐𝑐𝒌𝒌𝑧𝑧 ⁄𝜔𝜔 is the normalized
component of the 𝑘𝑘-vector along 𝑧𝑧 (i.e., the propagation direction), 𝜔𝜔 is the angular frequency of
the light wave, 𝑐𝑐 is the speed of light in vacuum, and ℏ is the reduced Planck constant.
Section 3 is devoted to an analysis of the energy and angular momentum of a quasi-static
system containing electrical charge and current. At first sight, the system appears to violate the
law of conservation of angular momentum. We pinpoint the cause of this discrepancy, and point
out the whereabouts of the missing angular momentum.
The case of a slowly rotating permanent magnet in an external electric field is taken up in
Sec.4. This quasi-static problem reveals the perils and pitfalls of ignoring the contribution of the
physical source of the 𝐸𝐸-field (i.e., a pair of electrically-charged parallel plates) when balancing
the mechanical momentum imparted to the magnet against the EM momentum contained in the
field. Additionally, the same problem, when analyzed in accordance with the Lorentz formalism,
shows the important contribution of “hidden momentum” to the physical behavior of the magnet.
In Sec.5 we examine the case of a slowly rotating electric dipole in an external magnetic
field. Here, once again, the physical source of the external field (i.e., a solenoid) turns out to be
of crucial significance when one attempts to verify the conservation of linear momentum.
Section 6 describes the connection between the energy of a dipole in an external EM field
and the energy content of the field in the region surrounding the dipole. Both cases of an electric
dipole in an external 𝐸𝐸-field and a magnetic dipole in an external 𝐻𝐻-field (or 𝐵𝐵-field) will be
considered. As before, the physical source of the external field will be seen to play an important
role when confirming the conservation of energy.
The final section contains a few concluding remarks and general observations. The integrals
needed throughout the paper are listed in Appendix A.

2. A light bullet described by Maxwell’s equations in cylindrical coordinates. Consider a

monochromatic wave of frequency 𝜔𝜔, propagating in free space along the z-axis. The wave-
number is 𝑘𝑘𝑧𝑧 = 𝑘𝑘0 𝜎𝜎𝑧𝑧 = (𝜔𝜔/𝑐𝑐)𝜎𝜎𝑧𝑧 , where 0 < 𝜎𝜎𝑧𝑧 < 1. The wave also exhibits an azimuthal phase

1
of 2𝜋𝜋𝜋𝜋 around the z-axis, where 𝑚𝑚, an integer, may be positive, zero, or negative. In cylindrical
coordinates, the electric field 𝑬𝑬(𝒓𝒓, 𝑡𝑡) and the magnetic field 𝑯𝑯(𝒓𝒓, 𝑡𝑡) of the wave are given by

𝑬𝑬(𝑟𝑟, 𝜑𝜑, 𝑧𝑧, 𝑡𝑡) = �𝐸𝐸𝑟𝑟 (𝑟𝑟)𝒓𝒓� + 𝐸𝐸𝜑𝜑 (𝑟𝑟)𝝋𝝋

� + 𝐸𝐸𝑧𝑧 (𝑟𝑟)𝒛𝒛��exp[i(𝑚𝑚𝑚𝑚 + 𝑘𝑘𝑧𝑧 𝑧𝑧 − 𝜔𝜔𝜔𝜔)], (1a)

𝑯𝑯(𝑟𝑟, 𝜑𝜑, 𝑧𝑧, 𝑡𝑡) = �𝐻𝐻𝑟𝑟 (𝑟𝑟)𝒓𝒓� + 𝐻𝐻𝜑𝜑 (𝑟𝑟)𝝋𝝋

� + 𝐻𝐻𝑧𝑧 (𝑟𝑟)𝒛𝒛��exp[i(𝑚𝑚𝑚𝑚 + 𝑘𝑘𝑧𝑧 𝑧𝑧 − 𝜔𝜔𝜔𝜔)]. (1b)

Maxwell’s equations [4-8] can now be used to relate the various components of the E and H
fields to each other, that is,
𝑟𝑟 −1 𝐸𝐸𝑟𝑟 + 𝐸𝐸𝑟𝑟′ + i𝑚𝑚𝑟𝑟 −1 𝐸𝐸𝜑𝜑 + i𝑘𝑘𝑧𝑧 𝐸𝐸𝑧𝑧 = 0, (2)

𝑚𝑚𝑟𝑟 −1 𝐸𝐸𝑧𝑧 − 𝑘𝑘𝑧𝑧 𝐸𝐸𝜑𝜑 = 𝜇𝜇0 𝜔𝜔𝐻𝐻𝑟𝑟 , (3a)

𝑘𝑘𝑧𝑧 𝐸𝐸𝑟𝑟 + i𝐸𝐸𝑧𝑧′ = 𝜇𝜇0 𝜔𝜔𝐻𝐻𝜑𝜑 , (3b)
i𝑟𝑟 −1 𝐸𝐸𝜑𝜑 + i𝐸𝐸𝜑𝜑′ + 𝑚𝑚𝑟𝑟 −1 𝐸𝐸𝑟𝑟 = −𝜇𝜇0 𝜔𝜔𝜔𝜔𝑧𝑧 , (3c)

𝑚𝑚𝑟𝑟 −1 𝐻𝐻𝑧𝑧 − 𝑘𝑘𝑧𝑧 𝐻𝐻𝜑𝜑 = −𝜀𝜀0 𝜔𝜔𝐸𝐸𝑟𝑟 , (4a)

𝑘𝑘𝑧𝑧 𝐻𝐻𝑟𝑟 + i𝐻𝐻𝑧𝑧′ = −𝜀𝜀0 𝜔𝜔𝐸𝐸𝜑𝜑 , (4b)
i𝑟𝑟 −1 𝐻𝐻𝜑𝜑 + i𝐻𝐻𝜑𝜑′ + 𝑚𝑚𝑟𝑟 −1 𝐻𝐻𝑟𝑟 = 𝜀𝜀0 𝜔𝜔𝐸𝐸𝑧𝑧 , (4c)

𝑟𝑟 −1 𝐻𝐻𝑟𝑟 + 𝐻𝐻𝑟𝑟′ + i𝑚𝑚𝑟𝑟 −1 𝐻𝐻𝜑𝜑 + i𝑘𝑘𝑧𝑧 𝐻𝐻𝑧𝑧 = 0. (5)

With the help of Eqs.(3a) and (3b), Eq.(4c) may be written as

i𝑟𝑟 −1 (𝑘𝑘𝑧𝑧 𝐸𝐸𝑟𝑟 + i𝐸𝐸𝑧𝑧′ ) + i(𝑘𝑘𝑧𝑧 𝐸𝐸𝑟𝑟′ + i𝐸𝐸𝑧𝑧″ ) + 𝑚𝑚𝑟𝑟 −1 �𝑚𝑚𝑟𝑟 −1 𝐸𝐸𝑧𝑧 − 𝑘𝑘𝑧𝑧 𝐸𝐸𝜑𝜑 � = (𝜔𝜔/𝑐𝑐)2 𝐸𝐸𝑧𝑧 .

Rearranging the terms of the above equation yields

i𝑘𝑘𝑧𝑧 �𝑟𝑟 −1 𝐸𝐸𝑟𝑟 + 𝐸𝐸𝑟𝑟′ + i𝑚𝑚𝑟𝑟 −1 𝐸𝐸𝜑𝜑 � − 𝐸𝐸𝑧𝑧″ − 𝑟𝑟 −1 𝐸𝐸𝑧𝑧′ + 𝑚𝑚2 𝑟𝑟 −2 𝐸𝐸𝑧𝑧 − (𝜔𝜔/𝑐𝑐)2 𝐸𝐸𝑧𝑧 = 0.
Invoking Eq.(2), we find
𝐸𝐸𝑧𝑧″ + 𝑟𝑟 −1 𝐸𝐸𝑧𝑧′ + [(𝜔𝜔/𝑐𝑐)2 − 𝑘𝑘𝑧𝑧2 − (𝑚𝑚/𝑟𝑟)2 ]𝐸𝐸𝑧𝑧 = 0. (6)
A similar operation, starting with Eq.(3c) and with the help of Eqs.(4a), (4b) and (5), yields
𝐻𝐻𝑧𝑧″ + 𝑟𝑟 −1 𝐻𝐻𝑧𝑧′ + [(𝜔𝜔/𝑐𝑐)2 − 𝑘𝑘𝑧𝑧2 − (𝑚𝑚/𝑟𝑟)2 ]𝐻𝐻𝑧𝑧 = 0. (7)
We thus find two sets of solutions to Maxwell’s equations, one with 𝐸𝐸𝑧𝑧 = 0, referred to as
transverse electric (TE), and another with 𝐻𝐻𝑧𝑧 = 0, known as transverse magnetic (TM). Defining
𝑘𝑘𝑟𝑟 = 𝑘𝑘0 𝜎𝜎𝑟𝑟 = (𝜔𝜔/𝑐𝑐)𝜎𝜎𝑟𝑟 , where 0 < 𝜎𝜎𝑟𝑟 < 1 and 𝜎𝜎𝑟𝑟2 + 𝜎𝜎𝑧𝑧2 = 1, the solutions to Eqs.(6) and (7) may
be written in terms of Bessel functions of the first kind, order 𝑚𝑚, as
𝐸𝐸𝑧𝑧 (𝑟𝑟) = 𝐸𝐸𝑧𝑧0 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟); (𝐻𝐻𝑧𝑧 = 0; TM), (8)
𝐻𝐻𝑧𝑧 (𝑟𝑟) = 𝐻𝐻𝑧𝑧0 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟); (𝐸𝐸𝑧𝑧 = 0; TE). (9)
In what follows, we will determine the remaining components of the E and H fields for the
two cases of TE and TM modes. In each case we also find the energy and angular momentum
contained in a finite-diameter, finite duration light pulse.

2
2.1. Transverse Electric (TE) modes: Setting 𝐸𝐸𝑧𝑧 = 0 in Eq.(3a), we find 𝐸𝐸𝜑𝜑 = −(𝜇𝜇0 𝜔𝜔/𝑘𝑘𝑧𝑧 )𝐻𝐻𝑟𝑟 .
Substitution into Eq.(4b) then yields
′ (𝑘𝑘
𝐻𝐻𝑟𝑟 (𝑟𝑟) = i(𝜎𝜎𝑧𝑧 /𝜎𝜎𝑟𝑟 )𝐻𝐻𝑧𝑧0 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟),
𝐻𝐻𝑟𝑟 (𝑟𝑟) = i𝑘𝑘𝑧𝑧 [(𝜔𝜔/𝑐𝑐)2 − 𝑘𝑘𝑧𝑧2 ]−1 𝐻𝐻𝑧𝑧′ (𝑟𝑟) → � ′
(8)
𝐸𝐸𝜑𝜑 (𝑟𝑟) = −i(𝑍𝑍0 /𝜎𝜎𝑟𝑟 )𝐻𝐻𝑧𝑧0 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟).

In the above equation, 𝑍𝑍0 = �𝜇𝜇0 /𝜀𝜀0 is the impedance of free space. Subsequently, Eq.(3c)
yields
𝐸𝐸𝑟𝑟 (𝑟𝑟) = −[𝑚𝑚𝑍𝑍0 /(𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟 𝑟𝑟)]𝐻𝐻𝑧𝑧0 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟). (9)
Similarly, from Eq.(4a) we find

𝐻𝐻𝜑𝜑 (𝑟𝑟) = −[𝑚𝑚𝑚𝑚𝑧𝑧 /(𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟 𝑟𝑟)]𝐻𝐻𝑧𝑧0 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟). (10)
The component of the Poynting vector along the z-axis is readily obtained as follows:
〈𝑆𝑆𝑧𝑧 (𝒓𝒓, 𝑡𝑡)〉 = ½Re�𝐸𝐸𝑟𝑟 𝐻𝐻𝜑𝜑∗ − 𝐸𝐸𝜑𝜑 𝐻𝐻𝑟𝑟∗ �
2 2 (𝑘𝑘 2 ′ 2
= ½[(𝑚𝑚2 𝜎𝜎𝑧𝑧 𝑍𝑍0 )/(𝑘𝑘𝑟𝑟 𝜎𝜎𝑟𝑟 𝑟𝑟)2 ]𝐻𝐻𝑧𝑧0 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟) + ½(𝑍𝑍0 𝜎𝜎𝑧𝑧 /𝜎𝜎𝑟𝑟2 )𝐻𝐻𝑧𝑧0 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)
2 2
= ½(𝑍𝑍0 𝐻𝐻𝑧𝑧0 𝜎𝜎𝑧𝑧 /𝜎𝜎𝑟𝑟2 ) �(𝑚𝑚/𝑘𝑘𝑟𝑟 𝑟𝑟)2 𝐽𝐽𝑚𝑚
2 (𝑘𝑘 ′
𝑟𝑟 𝑟𝑟) + 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)�. (11)

Integration over the cross-sectional area of the beam from 𝑟𝑟 = 0 to 𝑅𝑅 yields

𝑅𝑅 2 𝑅𝑅 ′ 2
∫0 2𝜋𝜋〈𝑆𝑆𝑧𝑧 (𝒓𝒓, 𝑡𝑡)〉𝑟𝑟𝑟𝑟𝑟𝑟 = (𝜋𝜋𝑍𝑍0 𝐻𝐻𝑧𝑧0 𝜎𝜎𝑧𝑧 /𝜎𝜎𝑟𝑟2 ) ∫0 𝑟𝑟�(𝑚𝑚/𝑘𝑘𝑟𝑟 𝑟𝑟)2 𝐽𝐽𝑚𝑚
2 (𝑘𝑘
𝑟𝑟 𝑟𝑟) + 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)�𝑑𝑑𝑑𝑑

2 𝑘𝑘 𝑅𝑅 2
= (𝜋𝜋𝑍𝑍0 𝐻𝐻𝑧𝑧0 𝜎𝜎𝑧𝑧 /𝑘𝑘𝑟𝑟2 𝜎𝜎𝑟𝑟2 ) ∫0 𝑟𝑟 𝑥𝑥�(𝑚𝑚/𝑥𝑥)2 𝐽𝐽𝑚𝑚
2 (𝑥𝑥) ′
+ 𝐽𝐽𝑚𝑚 (𝑥𝑥)�𝑑𝑑𝑑𝑑
2 2
= (½𝜋𝜋𝑅𝑅 2 𝑍𝑍0 𝐻𝐻𝑧𝑧0 𝜎𝜎𝑧𝑧 /𝜎𝜎𝑟𝑟2 ){𝐽𝐽𝑚𝑚+1 (𝑘𝑘𝑟𝑟 𝑅𝑅) − 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑅𝑅)𝐽𝐽𝑚𝑚+2 (𝑘𝑘𝑟𝑟 𝑅𝑅) + 2𝑚𝑚[𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑅𝑅)/(𝑘𝑘𝑟𝑟 𝑅𝑅)]2 }. (12)
Since Bessel beams in free space travel along the z-axis at the speed of 𝑐𝑐𝜎𝜎𝑧𝑧 , the integrated
Poynting vector in Eq.(12) should be equal to the EM energy contained in a cylindrical volume
of radius 𝑅𝑅 and length 𝑐𝑐𝜎𝜎𝑧𝑧 , where 𝑐𝑐 = 1/�𝜇𝜇0 𝜀𝜀0 is the speed of light in vacuum. Alternatively,
one could calculate the energy content of the cylindrical volume directly, using the total time-
averaged energy-density of the E and H fields, as follows:
〈ℰ(𝒓𝒓, 𝑡𝑡)〉 = ¼𝜀𝜀0 Re�𝐸𝐸𝑟𝑟 𝐸𝐸𝑟𝑟∗ + 𝐸𝐸𝜑𝜑 𝐸𝐸𝜑𝜑∗ � + ¼𝜇𝜇0 Re�𝐻𝐻𝑟𝑟 𝐻𝐻𝑟𝑟∗ + 𝐻𝐻𝜑𝜑 𝐻𝐻𝜑𝜑∗ + 𝐻𝐻𝑧𝑧 𝐻𝐻𝑧𝑧∗ �
2 2 (𝑘𝑘 2 ′ 2
= ¼𝜀𝜀0 [𝑚𝑚𝑍𝑍0 /(𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟 𝑟𝑟)]2 𝐻𝐻𝑧𝑧0 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟) + ¼𝜀𝜀0 (𝑍𝑍0 /𝜎𝜎𝑟𝑟 )2 𝐻𝐻𝑧𝑧0 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)
2 ′ 2 (𝑘𝑘 2 2 (𝑘𝑘 2 2
+¼𝜇𝜇0 (𝜎𝜎𝑧𝑧 /𝜎𝜎𝑟𝑟 )2 𝐻𝐻𝑧𝑧0 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟) + ¼𝜇𝜇0 [𝑚𝑚𝑚𝑚𝑧𝑧 /(𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟 𝑟𝑟)]2 𝐻𝐻𝑧𝑧0 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟) + ¼𝜇𝜇0 𝐻𝐻𝑧𝑧0 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)
2 2 (𝑘𝑘 2 ′ 2
= ¼𝜇𝜇0 [2𝑚𝑚2 /(𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟 𝑟𝑟)2 − (𝑚𝑚/𝑘𝑘𝑟𝑟 𝑟𝑟)2 + 1]𝐻𝐻𝑧𝑧0 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟) + ¼𝜇𝜇0 [(2/𝜎𝜎𝑟𝑟2 ) − 1]𝐻𝐻𝑧𝑧0 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)
2 2
= ½(𝜇𝜇0 𝐻𝐻𝑧𝑧0 /𝜎𝜎𝑟𝑟2 )�(𝑚𝑚/𝑘𝑘𝑟𝑟 𝑟𝑟)2 𝐽𝐽𝑚𝑚
2 (𝑘𝑘 ′
𝑟𝑟 𝑟𝑟) + 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)�
2 2 (𝑘𝑘 2 2 ′ 2
+¼𝜇𝜇0 𝐻𝐻𝑧𝑧0 �𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟) − (𝑚𝑚/𝑘𝑘𝑟𝑟 𝑟𝑟) 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟) − 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)�. (13)

In unit time, the Bessel beam propagates a distance 𝑐𝑐𝜎𝜎𝑧𝑧 along the z-axis. Multiplying 𝑐𝑐𝜎𝜎𝑧𝑧
into Eq.(13) yields a first term that is identical to Eq.(11). The second term, therefore, must

3
integrate to zero over the cross-sectional area of the beam. The integral of the second term on the
right-hand-side of Eq.(13) over the cylindrical volume of radius 𝑅𝑅 and length 𝑐𝑐𝜎𝜎𝑧𝑧 may be written
2 𝑅𝑅 2 (𝑘𝑘 ′ 2 ′
½𝜋𝜋𝜎𝜎𝑧𝑧 𝑍𝑍0 𝐻𝐻𝑧𝑧0 ∫0 { 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟) − [(𝑚𝑚/𝑘𝑘𝑟𝑟 𝑟𝑟) 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟) − 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)] − 2(𝑚𝑚/𝑘𝑘𝑟𝑟 𝑟𝑟) 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟) 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)} 𝑟𝑟𝑟𝑟𝑟𝑟

2 𝑘𝑘 𝑅𝑅 2
= ½𝜋𝜋(𝜎𝜎𝑧𝑧 /𝑘𝑘𝑟𝑟2 )𝑍𝑍0 𝐻𝐻𝑧𝑧0 𝑟𝑟 2 (𝑥𝑥)
∫0 [𝑥𝑥𝐽𝐽𝑚𝑚 − 𝑥𝑥𝐽𝐽𝑚𝑚+1 ′ (𝑥𝑥)]
(𝑥𝑥) − 2𝑚𝑚 𝐽𝐽𝑚𝑚 (𝑥𝑥) 𝐽𝐽𝑚𝑚 𝑑𝑑𝑑𝑑
2 [𝑘𝑘
= ½𝜋𝜋(𝜎𝜎𝑧𝑧 /𝑘𝑘𝑟𝑟2 )𝑍𝑍0 𝐻𝐻𝑧𝑧0 𝑟𝑟 𝑅𝑅 𝐽𝐽𝑚𝑚+1 (𝑘𝑘𝑟𝑟 𝑅𝑅) − 𝑚𝑚𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑅𝑅)] 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑅𝑅). (14)
The above expression vanishes when 𝑅𝑅 is chosen to correspond to one of the zeroes of
𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑅𝑅). The energy content of a cylindrical volume of radius 𝑅𝑅 and length 𝑐𝑐𝜎𝜎𝑧𝑧 is thus given by
Eq.(12) provided that 𝑅𝑅 is chosen such that 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑅𝑅) = 0.

Next we compute the angular momentum density of the TE mode, as follows:

〈𝐿𝐿𝑧𝑧 (𝒓𝒓, 𝑡𝑡)〉𝒛𝒛� = 𝑟𝑟𝒓𝒓� × ½Re(−𝐸𝐸𝑟𝑟 𝐻𝐻𝑧𝑧∗ 𝝋𝝋 2
�/𝑐𝑐 2 ) = ½[𝑚𝑚𝑍𝑍0 𝐻𝐻𝑧𝑧0 2 (𝑘𝑘
/(𝑐𝑐 2 𝑘𝑘𝑟𝑟 𝜎𝜎𝑟𝑟 )] 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟)𝒛𝒛
�. (15)
Integrating the above expression over the cross-sectional area of the beam from 𝑟𝑟 = 0 to 𝑅𝑅 yields
𝑅𝑅 2 𝑅𝑅
∫0 2𝜋𝜋𝜋𝜋〈𝐿𝐿𝑧𝑧 (𝒓𝒓, 𝑡𝑡)〉𝑑𝑑𝑑𝑑 = [𝜋𝜋𝜋𝜋𝑍𝑍0 𝐻𝐻𝑧𝑧0 2 (𝑘𝑘
/(𝑐𝑐 2 𝑘𝑘𝑟𝑟 𝜎𝜎𝑟𝑟 )] ∫0 𝑟𝑟 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟)𝑑𝑑𝑑𝑑

2 𝑘𝑘 𝑅𝑅
= [𝜋𝜋𝜋𝜋𝑍𝑍0 𝐻𝐻𝑧𝑧0 /(𝑐𝑐 2 𝑘𝑘𝑟𝑟3 𝜎𝜎𝑟𝑟 )] ∫0 𝑟𝑟 𝑥𝑥 𝐽𝐽𝑚𝑚
2 (𝑥𝑥)𝑑𝑑𝑑𝑑

2 )/(𝑐𝑐 2
= [½𝑚𝑚(𝜋𝜋𝑅𝑅 2 𝑍𝑍0 𝐻𝐻𝑧𝑧0 2 (𝑘𝑘
𝑘𝑘𝑟𝑟 𝜎𝜎𝑟𝑟 )][𝐽𝐽𝑚𝑚 𝑟𝑟 𝑅𝑅) − 𝐽𝐽𝑚𝑚−1 (𝑘𝑘𝑟𝑟 𝑅𝑅) 𝐽𝐽𝑚𝑚+1 (𝑘𝑘𝑟𝑟 𝑅𝑅)]
2 )/(𝑐𝑐𝑐𝑐𝜎𝜎 2 )][𝐽𝐽2 (𝑘𝑘
= [½𝑚𝑚(𝜋𝜋𝑅𝑅 2 𝑍𝑍0 𝐻𝐻𝑧𝑧0 𝑟𝑟 𝑚𝑚 𝑟𝑟 𝑅𝑅) − 𝐽𝐽𝑚𝑚−1 (𝑘𝑘𝑟𝑟 𝑅𝑅) 𝐽𝐽𝑚𝑚+1 (𝑘𝑘𝑟𝑟 𝑅𝑅)]. (16)

In unit time the beam propagates a distance of 𝑐𝑐𝜎𝜎𝑧𝑧 , so the angular momentum in Eq.(16)
must be multiplied by 𝑐𝑐𝜎𝜎𝑧𝑧 in order to correspond to the same amount of energy that is
represented by Eq.(12). The ratio of angular momentum to energy content of a cylindrical
volume of radius 𝑅𝑅 and arbitrary length, assuming that 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑅𝑅) = 0, is thus given by
−(𝑚𝑚⁄𝜔𝜔)𝐽𝐽𝑚𝑚−1 (𝑘𝑘𝑟𝑟 𝑅𝑅)/𝐽𝐽𝑚𝑚+1 (𝑘𝑘𝑟𝑟 𝑅𝑅). For sufficiently large 𝑅𝑅, the asymptotic value of 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑅𝑅) is
�2/(𝜋𝜋𝑘𝑘𝑟𝑟 𝑅𝑅) cos(𝑘𝑘𝑟𝑟 𝑅𝑅 − ½𝑚𝑚𝑚𝑚 − ¼𝜋𝜋), which leads to 𝐽𝐽𝑚𝑚−1 (𝑘𝑘𝑟𝑟 𝑅𝑅)/𝐽𝐽𝑚𝑚+1 (𝑘𝑘𝑟𝑟 𝑅𝑅) ≅ −1. Consequently,
the ratio of angular momentum to energy for a sufficiently large number of rings of the TE mode
of a Bessel beam in vacuum is given by 𝑚𝑚/𝜔𝜔. In quantum-optical terms, where the energy of
each photon is ℏ𝜔𝜔, the angular momentum per photon will be 𝑚𝑚ℏ.

2.2. Transverse Magnetic (TM) modes: Setting 𝐻𝐻𝑧𝑧 = 0 in Eq.(4a), we find 𝐻𝐻𝜑𝜑 = (𝜀𝜀0 𝜔𝜔/𝑘𝑘𝑧𝑧 )𝐸𝐸𝑟𝑟 .
Substitution into Eq.(3b) then yields
′ (𝑘𝑘
𝐸𝐸𝑟𝑟 (𝑟𝑟) = i(𝜎𝜎𝑧𝑧 /𝜎𝜎𝑟𝑟 )𝐸𝐸𝑧𝑧0 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟),
[(𝜔𝜔/𝑐𝑐) 2 2 ]−1 ′
𝐸𝐸𝑟𝑟 (𝑟𝑟) = i𝑘𝑘𝑧𝑧 − 𝑘𝑘𝑧𝑧 𝐸𝐸𝑧𝑧 (𝑟𝑟) → � ′ (𝑘𝑘
(17)
𝐻𝐻𝜑𝜑 (𝑟𝑟) = i(𝐸𝐸𝑧𝑧0 /𝑍𝑍0 𝜎𝜎𝑟𝑟 ) 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟).
Subsequently, Eq.(4c) yields
𝐻𝐻𝑟𝑟 (𝑟𝑟) = (𝑚𝑚𝐸𝐸𝑧𝑧0 /𝑍𝑍0 𝑘𝑘𝑟𝑟 𝜎𝜎𝑟𝑟 𝑟𝑟)𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟). (18)
Similarly, from Eq.(3a) we find
𝐸𝐸𝜑𝜑 (𝑟𝑟) = −(𝑚𝑚𝑚𝑚𝑧𝑧 /𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟 𝑟𝑟)𝐸𝐸𝑧𝑧0 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟). (19)
Having determined the various components of the E and H fields of the TM mode of a
Bessel beam in vacuum, we proceed to compute the Poynting vector along the z-axis, as follows:

4
 〈𝑆𝑆𝑧𝑧 (𝒓𝒓, 𝑡𝑡)〉 = ½Re�𝐸𝐸𝑟𝑟 𝐻𝐻𝜑𝜑∗ − 𝐸𝐸𝜑𝜑 𝐻𝐻𝑟𝑟∗ �
2 )/(𝑍𝑍 2 )] ′ 2 (𝑘𝑘 2
= ½[(𝜎𝜎𝑧𝑧 𝐸𝐸𝑧𝑧0 0 𝜎𝜎𝑟𝑟 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟) + ½(𝜎𝜎𝑧𝑧 𝐸𝐸𝑧𝑧0 2 (𝑘𝑘
/𝑍𝑍0 )[𝑚𝑚/(𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟 𝑟𝑟)]2 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟)
2 )/(𝑍𝑍 2 )] ′ 2 (𝑘𝑘
= ½[(𝜎𝜎𝑧𝑧 𝐸𝐸𝑧𝑧0 0 𝜎𝜎𝑟𝑟
2 (𝑘𝑘
�𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟) + (𝑚𝑚/𝑘𝑘𝑟𝑟 𝑟𝑟)2 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟)�. (20)
Integration over the cross-sectional area of the beam from 𝑟𝑟 = 0 to 𝑅𝑅 yields
𝑅𝑅 2
∫0 2𝜋𝜋𝜋𝜋〈𝑆𝑆𝑧𝑧 (𝒓𝒓, 𝑡𝑡)〉𝑑𝑑𝑑𝑑 = ½(𝜋𝜋𝑅𝑅 2 𝜎𝜎𝑧𝑧 𝐸𝐸𝑧𝑧0 /𝑍𝑍0 𝜎𝜎𝑟𝑟2 )
2 (𝑘𝑘𝑟𝑟 𝑅𝑅) − 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑅𝑅)𝐽𝐽𝑚𝑚+2 (𝑘𝑘𝑟𝑟 𝑅𝑅) + 2𝑚𝑚[𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑅𝑅)/(𝑘𝑘𝑟𝑟 𝑅𝑅)]2 }.
× { 𝐽𝐽𝑚𝑚+1 (21)
2
Aside from the coefficient 𝐸𝐸𝑧𝑧0 /𝑍𝑍0 , this result is identical to that obtained in Eq.(12) for a TE
2
beam, where the corresponding coefficient was 𝑍𝑍0 𝐻𝐻𝑧𝑧0 .

In similar fashion, the angular momentum density of a TM Bessel beam is found to be

〈𝐿𝐿𝑧𝑧 (𝒓𝒓, 𝑡𝑡)〉𝒛𝒛� = 𝑟𝑟𝒓𝒓� × ½Re(𝐸𝐸𝑧𝑧 𝐻𝐻𝑟𝑟∗ 𝝋𝝋 2
� /𝑐𝑐 2 ) = ½[𝑚𝑚𝐸𝐸𝑧𝑧0 2 (𝑘𝑘
/(𝑍𝑍0 𝑐𝑐 2 𝑘𝑘𝑟𝑟 𝜎𝜎𝑟𝑟 )] 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟)𝒛𝒛
�. (22)
Integration over the cross-sectional area of the beam from 𝑟𝑟 = 0 to 𝑅𝑅 yields
𝑅𝑅 2 𝑅𝑅
∫0 2𝜋𝜋𝜋𝜋〈𝐿𝐿𝑧𝑧 (𝒓𝒓, 𝑡𝑡)〉𝑑𝑑𝑑𝑑 = [𝜋𝜋𝜋𝜋𝐸𝐸𝑧𝑧0 2 (𝑘𝑘
/(𝑍𝑍0 𝑐𝑐 2 𝑘𝑘𝑟𝑟 𝜎𝜎𝑟𝑟 )] ∫0 𝑟𝑟 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟)𝑑𝑑𝑑𝑑

2 𝑘𝑘 𝑅𝑅
= [𝜋𝜋𝜋𝜋𝐸𝐸𝑧𝑧0 /(𝑍𝑍0 𝑐𝑐 2 𝑘𝑘𝑟𝑟3 𝜎𝜎𝑟𝑟 )] ∫0 𝑟𝑟 𝑥𝑥 𝐽𝐽𝑚𝑚
2 (𝑥𝑥)𝑑𝑑𝑑𝑑

2 )/(𝑍𝑍 2
= ½𝑚𝑚[(𝜋𝜋𝑅𝑅 2 𝐸𝐸𝑧𝑧0 2
0 𝑐𝑐 𝑘𝑘𝑟𝑟 𝜎𝜎𝑟𝑟 )][ 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑅𝑅) − 𝐽𝐽𝑚𝑚−1 (𝑘𝑘𝑟𝑟 𝑅𝑅) 𝐽𝐽𝑚𝑚+1 (𝑘𝑘𝑟𝑟 𝑅𝑅)]
2 )/(𝑍𝑍
= ½𝑚𝑚[(𝜋𝜋𝑅𝑅 2 𝐸𝐸𝑧𝑧0 2 2
0 𝑐𝑐𝑐𝑐𝜎𝜎𝑟𝑟 )][𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑅𝑅) − 𝐽𝐽𝑚𝑚−1 (𝑘𝑘𝑟𝑟 𝑅𝑅) 𝐽𝐽𝑚𝑚+1 (𝑘𝑘𝑟𝑟 𝑅𝑅)]. (23)

In unit time the beam propagates a distance of 𝑐𝑐𝜎𝜎𝑧𝑧 , so the angular momentum in Eq.(23)
must be multiplied by 𝑐𝑐𝜎𝜎𝑧𝑧 in order to correspond to the same amount of energy that is
represented by Eq.(21). Once again, the ratio of angular momentum to energy contained in a
cylindrical volume of radius 𝑅𝑅 and arbitrary length, where 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑅𝑅) = 0, is seen to approach
𝑚𝑚/𝜔𝜔 in the limit of large 𝑅𝑅. In quantum-optical terms, each photon of energy ℏ𝜔𝜔 in the TM
mode of the Bessel beam carries 𝑚𝑚ℏ of orbital angular momentum.

2.3. The case of circular polarization. We define a circularly-polarized Bessel beam as an equal
mixture of TE and TM modes, with the phase difference between the two eigen-modes being
±90° for right/left circular states. Thus, in the 𝑘𝑘-space, all plane-waves that constitute the
resulting Bessel beam will be circularly polarized. For a pair of TE and TM beams having the
same 𝜔𝜔, 𝑘𝑘𝑧𝑧 , and 𝑚𝑚, and also having 𝐸𝐸𝑧𝑧0 = ±i𝑍𝑍0 𝐻𝐻𝑧𝑧0, the superposed 𝑬𝑬 and 𝑯𝑯 fields are given by
′ (𝑘𝑘
𝐸𝐸𝑟𝑟 (𝑟𝑟) = ±i[𝑚𝑚𝐸𝐸𝑧𝑧0 /(𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟 𝑟𝑟)] 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟) + i(𝜎𝜎𝑧𝑧 /𝜎𝜎𝑟𝑟 )𝐸𝐸𝑧𝑧0 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟). (24)
′
𝐸𝐸𝜑𝜑 (𝑟𝑟) = ∓(𝐸𝐸𝑧𝑧0 /𝜎𝜎𝑟𝑟 ) 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟) − (𝑚𝑚𝑚𝑚𝑧𝑧 /𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟 𝑟𝑟)𝐸𝐸𝑧𝑧0 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟). (25)
𝐸𝐸𝑧𝑧 (𝑟𝑟) = 𝐸𝐸𝑧𝑧0 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟). (26)
′ (𝑘𝑘
𝐻𝐻𝑟𝑟 (𝑟𝑟) = ±(𝐸𝐸𝑧𝑧0 𝜎𝜎𝑧𝑧 /𝑍𝑍0 𝜎𝜎𝑟𝑟 ) 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟) + (𝑚𝑚𝐸𝐸𝑧𝑧0 /𝑍𝑍0 𝑘𝑘𝑟𝑟 𝜎𝜎𝑟𝑟 𝑟𝑟)𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟). (27)
′ (𝑘𝑘
𝐻𝐻𝜑𝜑 (𝑟𝑟) = ±i[𝑚𝑚𝑚𝑚𝑧𝑧 𝐸𝐸𝑧𝑧0 /(𝑍𝑍0 𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟 𝑟𝑟)] 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟) + i(𝐸𝐸𝑧𝑧0 /𝑍𝑍0 𝜎𝜎𝑟𝑟 ) 𝐽𝐽𝑚𝑚 𝑟𝑟 𝑟𝑟). (28)
𝐻𝐻𝑧𝑧 (𝑟𝑟) = ∓i(𝐸𝐸𝑧𝑧0 /𝑍𝑍0 ) 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟). (29)

5
The angular momentum density of this circularly-polarized Bessel beam is readily found to be
〈𝐿𝐿𝑧𝑧 (𝒓𝒓, 𝑡𝑡)〉𝒛𝒛� = 𝑟𝑟𝒓𝒓� × ½Re[(𝐸𝐸𝑧𝑧 𝐻𝐻𝑟𝑟∗ − 𝐸𝐸𝑟𝑟 𝐻𝐻𝑧𝑧∗ )𝝋𝝋
�/𝑐𝑐 2 ]
2
= [𝐸𝐸𝑧𝑧0 2 (𝑘𝑘
/(𝑍𝑍0 𝑐𝑐 2 𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟 )][𝑚𝑚𝐽𝐽𝑚𝑚 ′
𝑟𝑟 𝑟𝑟) ± 𝜎𝜎𝑧𝑧 𝑘𝑘𝑟𝑟 𝑟𝑟𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)]𝒛𝒛
�. (30)
Upon integration over the cross-sectional area of the beam (from 𝑟𝑟 = 0 to 𝑅𝑅) we find
𝑅𝑅
∫0 2𝜋𝜋𝜋𝜋〈𝐿𝐿𝑧𝑧 (𝒓𝒓, 𝑡𝑡)〉𝑑𝑑𝑑𝑑 =
2 𝑅𝑅
= [2𝜋𝜋𝐸𝐸𝑧𝑧0 2 (𝑘𝑘
/(𝑍𝑍0 𝑐𝑐 2 𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟 )] ∫0 [𝑚𝑚𝐽𝐽𝑚𝑚 ′
𝑟𝑟 𝑟𝑟) ± 𝜎𝜎𝑧𝑧 𝑘𝑘𝑟𝑟 𝑟𝑟𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)]𝑟𝑟𝑟𝑟𝑟𝑟

2 𝑘𝑘 𝑅𝑅
= [2𝜋𝜋𝐸𝐸𝑧𝑧0 /(𝑍𝑍0 𝑐𝑐 2 𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟3 )] ∫0 𝑟𝑟 [𝑚𝑚𝑚𝑚𝐽𝐽𝑚𝑚
2 (𝑥𝑥) ′ (𝑥𝑥)]𝑑𝑑𝑑𝑑
± 𝜎𝜎𝑧𝑧 𝑥𝑥 2 𝐽𝐽𝑚𝑚 (𝑥𝑥)𝐽𝐽𝑚𝑚
2
= [2𝜋𝜋𝐸𝐸𝑧𝑧0 /(𝑍𝑍0 𝑐𝑐 2 𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟3 )]
× {½𝑚𝑚(𝑘𝑘𝑟𝑟 𝑅𝑅)2 [𝐽𝐽𝑚𝑚
2 (𝑘𝑘 2
𝑟𝑟 𝑅𝑅) − 𝐽𝐽𝑚𝑚−1 (𝑘𝑘𝑟𝑟 𝑅𝑅)𝐽𝐽𝑚𝑚+1 (𝑘𝑘𝑟𝑟 𝑅𝑅)] ± ½𝜎𝜎𝑧𝑧 (𝑘𝑘𝑟𝑟 𝑅𝑅) 𝐽𝐽𝑚𝑚−1 (𝑘𝑘𝑟𝑟 𝑅𝑅)𝐽𝐽𝑚𝑚+1 (𝑘𝑘𝑟𝑟 𝑅𝑅)}

2
= [𝜋𝜋𝑅𝑅 2 𝐸𝐸𝑧𝑧0 /(𝑍𝑍0 𝑐𝑐𝑐𝑐𝜎𝜎𝑟𝑟2 )][(±𝜎𝜎𝑧𝑧 − 𝑚𝑚)𝐽𝐽𝑚𝑚−1 (𝑘𝑘𝑟𝑟 𝑅𝑅)𝐽𝐽𝑚𝑚+1 (𝑘𝑘𝑟𝑟 𝑅𝑅) + 𝑚𝑚𝐽𝐽𝑚𝑚
2 (𝑘𝑘
𝑟𝑟 𝑅𝑅)]. (31)
If 𝑅𝑅 is chosen to correspond to a zero of 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑅𝑅), the last term in Eq.(31) vanishes, and the
remaining terms give the angular momentum due to both vorticity and circular polarization.
Finally, we compute the z-component of the Poynting vector of the circularly-polarized
Bessel beam, as follows:
〈𝑆𝑆𝑧𝑧 (𝒓𝒓, 𝑡𝑡)〉 = ½Re�𝐸𝐸𝑟𝑟 𝐻𝐻𝜑𝜑∗ − 𝐸𝐸𝜑𝜑 𝐻𝐻𝑟𝑟∗ �
2 2
= ½{[𝑚𝑚2 𝜎𝜎𝑧𝑧 𝐸𝐸𝑧𝑧0 2 (𝑘𝑘
/𝑍𝑍0 (𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟 𝑟𝑟)2 ] 𝐽𝐽𝑚𝑚 2 ′
𝑟𝑟 𝑟𝑟) ± [𝑚𝑚𝐸𝐸𝑧𝑧0 /(𝑍𝑍0 𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟 𝑟𝑟)] 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟) 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)
2 2 2 ′ 2
±[𝑚𝑚𝑚𝑚𝑧𝑧2 𝐸𝐸𝑧𝑧0 /(𝑍𝑍0 𝜎𝜎𝑟𝑟2 𝑘𝑘𝑟𝑟 𝑟𝑟)] 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)𝐽𝐽𝑚𝑚
′ (𝑘𝑘
𝑟𝑟 𝑟𝑟) + (𝜎𝜎𝑧𝑧 𝐸𝐸𝑧𝑧0 /𝑍𝑍0 𝜎𝜎𝑟𝑟 )𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)
2 ′ 2 (𝑘𝑘 2
+ (𝜎𝜎𝑧𝑧 𝐸𝐸𝑧𝑧0 /𝑍𝑍0 𝜎𝜎𝑟𝑟2 ) 𝐽𝐽𝑚𝑚 2 ′
𝑟𝑟 𝑟𝑟) ± (𝑚𝑚𝐸𝐸𝑧𝑧0 /𝑍𝑍0 𝑘𝑘𝑟𝑟 𝜎𝜎𝑟𝑟 𝑟𝑟)𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)
2 2
±(𝑚𝑚𝜎𝜎𝑧𝑧2 𝐸𝐸𝑧𝑧0 /𝑍𝑍0 𝜎𝜎𝑟𝑟2 𝑘𝑘𝑟𝑟 𝑟𝑟)𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)𝐽𝐽𝑚𝑚
′ (𝑘𝑘 2 2 2
𝑟𝑟 𝑟𝑟) + [𝑚𝑚 𝜎𝜎𝑧𝑧 𝐸𝐸𝑧𝑧0 /𝑍𝑍0 (𝜎𝜎𝑟𝑟 𝑘𝑘𝑟𝑟 𝑟𝑟) ] 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)}

2 ′ 2
= (𝜎𝜎𝑧𝑧 𝐸𝐸𝑧𝑧0 /𝑍𝑍0 𝜎𝜎𝑟𝑟2 )�(𝑚𝑚/𝑘𝑘𝑟𝑟 𝑟𝑟)2 𝐽𝐽𝑚𝑚
2 (𝑘𝑘
𝑟𝑟 𝑟𝑟) + 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)�
2
±[𝑚𝑚(1 + 𝜎𝜎𝑧𝑧2 )𝐸𝐸𝑧𝑧0 /(𝑍𝑍0 𝜎𝜎𝑟𝑟2 𝑘𝑘𝑟𝑟 𝑟𝑟)] 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑟𝑟)𝐽𝐽𝑚𝑚
′ (𝑘𝑘
𝑟𝑟 𝑟𝑟). (32)
The first term on the right-hand-side of the above equation is twice the expression in
Eq.(20). Therefore, upon integration, this term yields the total power of the beam, as before. The
second term integrates to
2 𝑘𝑘 𝑅𝑅 2
±[2𝜋𝜋𝜋𝜋(1 + 𝜎𝜎𝑧𝑧2 )𝐸𝐸𝑧𝑧0 /(𝑍𝑍0 𝜎𝜎𝑟𝑟2 𝑘𝑘𝑟𝑟2 )] ∫0 𝑟𝑟 𝐽𝐽𝑚𝑚 (𝑥𝑥)𝐽𝐽𝑚𝑚
′ (𝑥𝑥)𝑑𝑑𝑑𝑑
= ±[𝜋𝜋𝜋𝜋(1 + 𝜎𝜎𝑧𝑧2 )𝐸𝐸𝑧𝑧0 /(𝑍𝑍0 𝜎𝜎𝑟𝑟2 𝑘𝑘𝑟𝑟2 )] 𝐽𝐽𝑚𝑚
2 (𝑘𝑘
𝑟𝑟 𝑅𝑅). (33)

This function vanishes if 𝑅𝑅 is chosen to coincide with a zero of 𝐽𝐽𝑚𝑚 (𝑘𝑘𝑟𝑟 𝑅𝑅). Thus the energy
content of our circularly-polarized Bessel beam is equal to the sum of the energies of its TE and
TM components. The ratio of angular momentum to energy (in the limit of large 𝑅𝑅) is now seen
to be (𝑚𝑚 ∓ 𝜎𝜎𝑧𝑧 )/𝜔𝜔, which includes the contribution ±𝜎𝜎𝑧𝑧 /𝜔𝜔 of spin to the orbital angular
momentum 𝑚𝑚/𝜔𝜔. The ± sign is related to the sense of circular polarization, which could be the
same as or opposite to the sense of the orbital angular momentum of the Bessel beam of order 𝑚𝑚.
When the Bessel beam has a large diameter (compared to a wavelength 𝜆𝜆 = 2𝜋𝜋𝜋𝜋/𝜔𝜔), its 𝜎𝜎𝑧𝑧 will
be close to 1.0, in which case the spin angular momentum per photon will be nearly ±ℏ. For
more compact Bessel beams, however, the spin angular momentum per photon will be ±ℏ𝜎𝜎𝑧𝑧 .

6
3. Electromagnetic energy and angular momentum of a rotating cylinder. As a second
example of simple EM systems, consider an infinitely long, uniformly-charged, non-conducting,
hollow cylinder of radius 𝑅𝑅, having a
surface-charge-density 𝜎𝜎𝑠𝑠 , as shown in (a) z (b) z
Fig.1(a). The E-field inside the cylinder is
zero, while that outside may be evaluated R R
using Gauss’s law [4,5], as follows: R1
− −
+ + + − − +
𝑬𝑬0 (𝒓𝒓) = 𝑅𝑅𝜎𝜎𝑠𝑠 𝒓𝒓�⁄(𝜀𝜀0 𝑟𝑟) ; 𝑟𝑟 > 𝑅𝑅. (34) + Ω0 − + −
− −
+ + + − − +
Fig.1. (a) Infinitely-long hollow cylinder of radius + − + −
𝑅𝑅, having a uniform surface-charge-density 𝜎𝜎𝑠𝑠 , is − −
+ + + − − +
brought from rest to a constant angular velocity Ω0
σ + − + −
+ −
around the 𝑧𝑧-axis. (b) A second infinitely-long s
− +
+ +
cylinder of radius 𝑅𝑅1 and uniform surface-charge-
density 𝜎𝜎𝑠𝑠1 resides inside the cylinder of radius 𝑅𝑅.
+ +
The inner cylinder, which does not rotate, is
concentric with the outer (rotating) cylinder.

Let the cylinder start to spin slowly around its axis, so that its angular velocity is brought
from zero to some finite value Ω0 after a long time, 𝑡𝑡max . During the time interval (0, 𝑡𝑡max ),
when the angular velocity is Ω(𝑡𝑡), the surface-current-density will be 𝑱𝑱𝑠𝑠 (𝑡𝑡) = 𝜎𝜎𝑠𝑠 𝑅𝑅Ω(𝑡𝑡)𝝋𝝋
� . Ignoring
the retardation effects (justified by the gradual rise of the spinning velocity), the vector potential
𝑨𝑨(𝒓𝒓, 𝑡𝑡) may be calculated as follows [4-8]:
𝜋𝜋
𝜇𝜇0 𝑱𝑱𝑠𝑠 (𝒓𝒓′ ,𝑡𝑡) 𝜇𝜇0 ∞ 2𝜎𝜎𝑠𝑠 𝑅𝑅Ω cos 𝜑𝜑
𝑨𝑨(𝒓𝒓, 𝑡𝑡) = � |𝒓𝒓 − 𝒓𝒓′ |
𝑑𝑑𝑠𝑠 ′ = 4𝜋𝜋 � �𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅
∫𝑧𝑧=−∞ �(𝑟𝑟 2 +𝑅𝑅2 −2𝑟𝑟𝑟𝑟 cos 𝜑𝜑)+𝑧𝑧 2 𝝋𝝋
4𝜋𝜋
𝜑𝜑=0
𝜋𝜋
𝜇𝜇0 𝜎𝜎𝑠𝑠 𝑅𝑅2 Ω 𝝋𝝋
� �(𝑟𝑟 2 +𝑅𝑅2 −2𝑟𝑟𝑟𝑟 cos 𝜑𝜑)+𝑧𝑧 2 +𝑧𝑧
= � cos 𝜑𝜑 lim𝑧𝑧→∞ �ln � 𝑑𝑑𝑑𝑑 See Appendix A
2𝜋𝜋 �(𝑟𝑟 2 +𝑅𝑅2 −2𝑟𝑟𝑟𝑟 cos 𝜑𝜑)+𝑧𝑧 2 −𝑧𝑧
0
𝜋𝜋 2
𝜇𝜇0 𝜎𝜎𝑠𝑠 𝑅𝑅2 Ω𝝋𝝋
� ��(𝑟𝑟 2 +𝑅𝑅2 −2𝑟𝑟𝑟𝑟 cos 𝜑𝜑)+𝑧𝑧 2 +𝑧𝑧�
= � cos 𝜑𝜑 lim𝑧𝑧→∞ �ln � 𝑑𝑑𝑑𝑑
2𝜋𝜋 𝑟𝑟 2 +𝑅𝑅2 −2𝑟𝑟𝑟𝑟 cos 𝜑𝜑
0
𝜋𝜋
𝜇𝜇0 𝜎𝜎𝑠𝑠 𝑅𝑅2 Ω𝝋𝝋
� 2 2 2
= 2𝜋𝜋
� cos 𝜑𝜑 lim �2 ln��(𝑟𝑟 + 𝑅𝑅 − 2𝑟𝑟𝑟𝑟 cos 𝜑𝜑) + 𝑧𝑧 + 𝑧𝑧�� 𝑑𝑑𝑑𝑑
0 𝑧𝑧→∞

𝜇𝜇0 𝜎𝜎𝑠𝑠 𝑅𝑅2 Ω𝝋𝝋

� 𝜋𝜋
− 2𝜋𝜋
∫0 cos 𝜑𝜑 ln(𝑟𝑟 2 + 𝑅𝑅 2 − 2𝑟𝑟𝑟𝑟 cos 𝜑𝜑) 𝑑𝑑𝑑𝑑
𝜋𝜋
𝜇𝜇0 𝜎𝜎𝑠𝑠 𝑅𝑅2 Ω𝝋𝝋
�
= 𝜋𝜋
� cos 𝜑𝜑 lim �ln(𝑧𝑧) + ln�1 + �1 + (𝑟𝑟 2 + 𝑅𝑅 2 − 2𝑟𝑟𝑟𝑟 cos 𝜑𝜑)/𝑧𝑧 2 �� 𝑑𝑑𝑑𝑑
0 𝑧𝑧→∞

𝜇𝜇0 𝜎𝜎𝑠𝑠 𝑅𝑅 2 Ω𝝋𝝋

� 𝜋𝜋
− 2𝜋𝜋
∫0 cos 𝜑𝜑 {ln(𝑅𝑅 2 ) + ln[1 − 2(𝑟𝑟/𝑅𝑅) cos 𝜑𝜑 + (𝑟𝑟/𝑅𝑅)2 ]}𝑑𝑑𝑑𝑑
𝜇𝜇0 𝜎𝜎𝑠𝑠 𝑅𝑅2 Ω𝝋𝝋
� 𝜋𝜋
= [lim𝑧𝑧→∞ ln(𝑧𝑧) + ln(2) − ln(𝑅𝑅)] ∫0 cos 𝜑𝜑 𝑑𝑑𝑑𝑑
𝜋𝜋
𝜇𝜇0 𝜎𝜎𝑠𝑠 𝑅𝑅2 Ω𝝋𝝋
� 𝜋𝜋
− 2𝜋𝜋
∫0 cos 𝜑𝜑 ln[1 − 2(𝑟𝑟/𝑅𝑅) cos 𝜑𝜑 + (𝑟𝑟/𝑅𝑅)2 ] 𝑑𝑑𝑑𝑑 See Appendix A

7
 �,
½(𝜇𝜇0 𝜎𝜎𝑠𝑠 𝑅𝑅Ω𝑟𝑟)𝝋𝝋 𝑟𝑟 < 𝑅𝑅;
=� (35)
½(𝜇𝜇0 𝜎𝜎𝑠𝑠 𝑅𝑅 3 Ω⁄𝑟𝑟)𝝋𝝋
�, 𝑟𝑟 > 𝑅𝑅.

The magnetic field produced by the spinning cylinder is thus found to be

𝜇𝜇0 𝜎𝜎𝑠𝑠 𝑅𝑅Ω(𝑡𝑡)𝒛𝒛� 𝑟𝑟 < 𝑅𝑅,
𝑩𝑩(𝒓𝒓, 𝑡𝑡) = 𝜇𝜇0 𝑯𝑯(𝒓𝒓, 𝑡𝑡) = 𝜵𝜵 × 𝑨𝑨(𝒓𝒓, 𝑡𝑡) = 𝑟𝑟 −1 𝜕𝜕𝑟𝑟 �𝑟𝑟𝐴𝐴𝜑𝜑 �𝒛𝒛� = � (36)
0 𝑟𝑟 > 𝑅𝑅.
While the cylinder spins up, an electric field is induced by the time-varying vector potential.
The magnitude of this 𝐸𝐸-field at the cylinder surface is
𝑬𝑬(𝑟𝑟 = 𝑅𝑅, 𝑡𝑡) = −𝜕𝜕𝑡𝑡 𝑨𝑨(𝑟𝑟 = 𝑅𝑅, 𝑡𝑡) = −½(𝜇𝜇0 𝜎𝜎𝑠𝑠 𝑅𝑅 2 Ω′ )𝝋𝝋
�. (37)
The mechanical energy (per unit length along the z-axis) needed to bring the cylinder from
rest to its final spinning speed Ω0 = Ω(𝑡𝑡max ) is obtained by integrating 𝑬𝑬 ∙ 𝑱𝑱𝑠𝑠 over the cylindrical
surface and over time from 𝑡𝑡 = 0 to 𝑡𝑡max . We will have
𝑡𝑡
ℰ = 2𝜋𝜋𝜋𝜋 ∫0 max 𝐸𝐸(𝑟𝑟 = 𝑅𝑅, 𝑡𝑡)𝜎𝜎𝑠𝑠 𝑅𝑅Ω(𝑡𝑡)𝑑𝑑𝑑𝑑
𝑡𝑡
= 𝜇𝜇0 𝜋𝜋𝑅𝑅 4 𝜎𝜎𝑠𝑠2 ∫0 max Ω′ (𝑡𝑡)Ω(𝑡𝑡)𝑑𝑑𝑑𝑑 = ½𝜇𝜇0 𝜋𝜋𝑅𝑅 4 𝜎𝜎𝑠𝑠2 Ω20 . (38)
This energy is stored in the internal B-field (or 𝐻𝐻-field) of the cylinder, whose density is
½𝜇𝜇0 𝐻𝐻 2 = ½𝜇𝜇0 (𝜎𝜎𝑠𝑠 𝑅𝑅Ω0 )2.

3.1. Angular momentum conservation. In the system of Fig.1(a), the torque (per unit length
along the z-axis) exerted on the cylinder is obtained by integrating 𝒓𝒓 × 𝜎𝜎𝑠𝑠 𝑬𝑬(𝑟𝑟 = 𝑅𝑅, 𝑡𝑡) over the
cylindrical surface, that is,
2𝜋𝜋
𝑻𝑻(𝑡𝑡) = ∫𝜑𝜑=0 𝑅𝑅𝒓𝒓� × (−½𝜇𝜇0 𝜎𝜎𝑠𝑠2 𝑅𝑅 2 Ω′ )𝝋𝝋
�𝑅𝑅𝑅𝑅𝑅𝑅 = −𝜋𝜋𝜇𝜇0 𝜎𝜎𝑠𝑠2 𝑅𝑅 4 Ω′ (𝑡𝑡)𝒛𝒛�. (39)
The total angular momentum picked up by the agency in charge of spinning the cylinder is
thus given by
𝑡𝑡
𝑳𝑳 = ∫0 max 𝑻𝑻(𝑡𝑡)𝑑𝑑𝑑𝑑 = −𝜋𝜋𝜇𝜇0 𝜎𝜎𝑠𝑠2 𝑅𝑅 4 Ω0 𝒛𝒛�. (40)
At this point, we are faced with a dilemma. In its final state, the charged cylinder is spinning
at the constant angular velocity Ω0 , producing a constant, uniform magnetic field 𝑯𝑯 = 𝜎𝜎𝑠𝑠 𝑅𝑅Ω0 𝒛𝒛�
inside the cylinder, and a constant, radially-decaying electric field 𝑬𝑬0 (𝑟𝑟) given by Eq.(34)
outside the cylinder. Therefore, the EM field produced by the spinning charged cylinder appears
to not have any angular momentum. In other words, the existence of the mechanical angular
momentum 𝑳𝑳 of Eq.(40), which is acquired by the spinning agent, appears to contradict the
conservation of angular momentum. This discrepancy, however, is easily resolved if one
recognizes the assumed infinite length of the cylinder as the culprit. A finite-length cylinder will
always produce a magnetic field in faraway regions (𝑟𝑟 ≫ 𝑅𝑅), and the EM angular momentum
density 𝓛𝓛𝑒𝑒𝑒𝑒 (𝒓𝒓) = 𝒓𝒓 × (𝑬𝑬 × 𝑯𝑯/𝑐𝑐 2 ), when integrated over the entire space, turns out to be exactly
equal in magnitude and opposite in direction to the mechanical angular momentum 𝑳𝑳 of Eq.(40).
A more tractable version of the above problem involves a second charged cylinder of radius
𝑅𝑅1 < 𝑅𝑅 placed inside (and concentric with) the spinning cylinder, as shown in Fig.1(b). If the
surface-charge-density of the inner cylinder is taken to be 𝜎𝜎𝑠𝑠1 = −𝑅𝑅𝜎𝜎𝑠𝑠 /𝑅𝑅1 , the static 𝐸𝐸-field will
be confined to the region between the two cylinders, as follows:

8
 0, 𝑟𝑟 < 𝑅𝑅1 ,
𝑬𝑬0 (𝒓𝒓) = �−𝑅𝑅𝜎𝜎𝑠𝑠 𝒓𝒓�/(𝜀𝜀0 𝑟𝑟), 𝑅𝑅1 < 𝑟𝑟 < 𝑅𝑅, (41)
0, 𝑟𝑟 > 𝑅𝑅.
The EM angular momentum (per unit length along z), which is now confined to the region
between the two cylinders, is given by
2𝜋𝜋 𝑅𝑅
𝑳𝑳𝑒𝑒𝑒𝑒 = ∫𝜑𝜑=0 ∫𝑟𝑟=𝑅𝑅 𝒓𝒓 × (𝑬𝑬0 × 𝑯𝑯/𝑐𝑐 2 )𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 = 𝜋𝜋𝜇𝜇0 𝜎𝜎𝑠𝑠2 𝑅𝑅 2 (𝑅𝑅 2 − 𝑅𝑅12 )Ω0 𝒛𝒛�. (42)
1

The difference between the mechanical angular momentum 𝑳𝑳 in Eq.(40) and the EM angular
momentum in Eq.(42) is taken up by the external agency responsible for holding the inner
cylinder stationary while the outer cylinder is spun from rest to its final angular velocity Ω0 .
Needless to say, no energy is spent in keeping the inner cylinder stationary. However, the
induced 𝐸𝐸-field 𝑬𝑬(𝑟𝑟 = 𝑅𝑅1 , 𝑡𝑡) = −𝜕𝜕𝑡𝑡 𝑨𝑨(𝑟𝑟 = 𝑅𝑅1 , 𝑡𝑡) = −½(𝜇𝜇0 𝜎𝜎𝑠𝑠 𝑅𝑅𝑅𝑅1 Ω′ )𝝋𝝋
� exerts a torque on the
inner cylinder, which, upon integration over time, yields
𝑡𝑡 2𝜋𝜋
∫𝜑𝜑=0 𝑅𝑅1 𝒓𝒓� × 𝜎𝜎𝑠𝑠1 𝑬𝑬(𝑟𝑟 = 𝑅𝑅1 , 𝑡𝑡)𝑅𝑅1 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 = 𝜋𝜋𝜇𝜇0 𝜎𝜎𝑠𝑠2 𝑅𝑅 2 𝑅𝑅12 Ω0 𝒛𝒛�.
max
𝑳𝑳1 = ∫𝑡𝑡=0 (43)
The total mechanical angular momentum 𝑳𝑳 + 𝑳𝑳1 is thus seen to be exactly equal in
magnitude and opposite in direction to the EM angular momentum 𝑳𝑳𝑒𝑒𝑒𝑒 stored in the 𝑬𝑬 and 𝑯𝑯
fields in the region between the two cylinders.
A similar argument can be made if the oppositely-charged stationary cylinder has a radius
𝑅𝑅1 > 𝑅𝑅. The EM momentum now vanishes, and the angular momenta of the cylinders cancel out.

3.2. Spinning cylinder in an external electric field. Next, we assume that the rotating cylinder
�, as shown in Fig.2. The charges on
of Fig.1(a) is placed inside a uniform electric field 𝑬𝑬 = 𝐸𝐸𝑥𝑥 𝒙𝒙
the cylinder and those on the plates attract/repel each other along the x-axis. However, no other
net force is exerted on the cylinder as it spins up from Ω = 0 to Ω0 . In contrast, the induced 𝐸𝐸-
field during the time-interval (0, 𝑡𝑡max ), namely,
𝑬𝑬(𝑟𝑟, 𝑡𝑡) = −𝜕𝜕𝑡𝑡 𝑨𝑨(𝑟𝑟, 𝑡𝑡) = −½(𝜇𝜇0 𝜎𝜎𝑠𝑠 𝑅𝑅 3 Ω′ /𝑟𝑟)𝝋𝝋
�, (44)
exerts a force in the y-direction on both plates. With reference to Fig.2 and considering that
𝑦𝑦 = (𝑑𝑑⁄2) tan 𝜑𝜑, the force (per unit-length along z) exerted on each plate is readily seen to be
½𝜋𝜋 𝑑𝑑
𝐹𝐹𝑦𝑦 (𝑡𝑡) = ∫𝜑𝜑=−½𝜋𝜋 −𝜎𝜎0 𝐸𝐸𝜑𝜑 �𝑟𝑟 = 2 cos 𝜑𝜑 , 𝑡𝑡� cos 𝜑𝜑 (𝑑𝑑 ⁄2)(1 + tan2 𝜑𝜑)𝑑𝑑𝑑𝑑 = ½𝜋𝜋𝜇𝜇0 𝜎𝜎0 𝜎𝜎𝑠𝑠 𝑅𝑅 3 Ω′ . (45)

dy y
+ −
+ σs −
+ Ω r
Fig.2. An infinitely-long, uniformly-charged cylinder of radius −
+
𝑅𝑅 and surface-charge-density 𝜎𝜎𝑠𝑠 rotates at an angular velocity Ω + ϕ −
around the z-axis. A pair of uniformly-charged parallel plates · x
produces a constant, uniform E-field in the cross-sectional plane + σ0 z −σ 0 −
of the cylinder. The surface-charge-density of the parallel plates + −
is ±𝜎𝜎0 , the E-field produced in the space between the plates is +
E −
𝑬𝑬 = (𝜎𝜎0 /𝜀𝜀0 )𝒙𝒙
�, the H-field inside the cylinder is 𝑯𝑯 = 𝜎𝜎𝑠𝑠 𝑅𝑅Ω 𝒛𝒛�, +
and the EM momentum-density within the cylinder is given by + −
𝓹𝓹𝑒𝑒𝑒𝑒 = 𝑬𝑬 × 𝑯𝑯/𝑐𝑐 2 = −(𝜇𝜇0 𝜎𝜎0 𝜎𝜎𝑠𝑠 𝑅𝑅Ω) 𝒚𝒚
�. d

9
 When the above force is integrated over time from 𝑡𝑡 = 0 to 𝑡𝑡max and multiplied by two (to
account for both plates), we find the mechanical momentum acquired by the plates at 𝑡𝑡 = 𝑡𝑡max to
be 𝒑𝒑 = 𝜋𝜋𝜇𝜇0 𝜎𝜎0 𝜎𝜎𝑠𝑠 𝑅𝑅 3 Ω0 𝒚𝒚
�. Considering that the E-field produced by the plates is 𝑬𝑬 = (𝜎𝜎0 /𝜀𝜀0 )𝒙𝒙
�,
and the H-field inside the cylinder is 𝑯𝑯 = 𝜎𝜎𝑠𝑠 𝑅𝑅Ω0 𝒛𝒛�, the EM momentum-density within the
cylinder will be 𝓹𝓹𝑒𝑒𝑒𝑒 = 𝑬𝑬 × 𝑯𝑯/𝑐𝑐 2 = −(𝜇𝜇0 𝜎𝜎0 𝜎𝜎𝑠𝑠 𝑅𝑅Ω0 )𝒚𝒚
�. Multiplication by the cross-sectional area
of the cylinder then shows that the EM momentum stored within the cylinder is precisely equal
in magnitude and opposite in sign to the mechanical momentum acquired by the plates.

3.3. Hidden momentum of the spinning cylinder. It has been argued that a stationary system
such as that of the spinning cylinder residing between the parallel plates of Fig.2 must have an
internal “hidden momentum” equal and opposite to its EM momentum, so that the net linear
momentum of the stationary system would be zero [9-32]. The acquisition of hidden momentum
by the cylinder thus dictates the transfer of a compensatory mechanical momentum to the
cylinder as it spins up from rest (Ω = 0) to its final state where Ω = Ω0 . The linear mechanical
momentum per unit-length of the cylinder at 𝑡𝑡 = 𝑡𝑡max is thus argued to be 𝒑𝒑mech =
−𝜋𝜋𝜇𝜇0 𝜎𝜎0 𝜎𝜎𝑠𝑠 𝑅𝑅 3 Ω0 𝒚𝒚
�. In this way, the system acquires no net mechanical momentum, as the
cylinder and the plates end up moving in opposite directions with equal momentum, while the
EM momentum is compensated by an equal and opposite amount of hidden momentum.
If the spinning cylinder happens to be used as a model for a uniformly-magnetized solid
cylinder, then two alternative formulations of classical electrodynamics may be used to describe
the situation discussed above. In the Lorentz formalism [4,5], the magnetization 𝑀𝑀0 𝒛𝒛� of the
cylinder is equivalent to a current-density 𝑱𝑱 = 𝜇𝜇0−1 𝜵𝜵 × 𝑴𝑴, which turns out to be the surface
current-density 𝑱𝑱𝑠𝑠 = 𝜇𝜇0−1 𝑀𝑀0 𝝋𝝋�. The H-field inside the cylinder is now zero, but the internal B-
field continues to be given by 𝑩𝑩 = 𝜇𝜇0 𝐽𝐽𝑠𝑠 𝒛𝒛�, which is equal to 𝑀𝑀0 𝒛𝒛�. In the Lorentz formalism, the
EM momentum-density is 𝓹𝓹𝑒𝑒𝑒𝑒 = 𝜀𝜀0 𝑬𝑬 × 𝑩𝑩 (known as Livens momentum), whereas the hidden
momentum-density is 𝓹𝓹hidden = 𝑴𝑴 × 𝜀𝜀0 𝑬𝑬. Therefore, all the preceding arguments remain intact.
In the Einstein-Laub formalism [33], where the EM momentum-density is 𝓹𝓹𝑒𝑒𝑒𝑒 = 𝑬𝑬 × 𝑯𝑯/𝑐𝑐 2,
there is no need for hidden momentum. In this case, the EM momentum of the system is zero.
However, the force-density exerted by an external E-field on the magnetization 𝑴𝑴(𝒓𝒓, 𝑡𝑡) is given
by 𝒇𝒇(𝒓𝒓, 𝑡𝑡) = −𝜕𝜕𝑡𝑡 𝑴𝑴 × 𝜀𝜀0 𝑬𝑬. If 𝑴𝑴 rises slowly from zero to 𝑀𝑀0 𝒛𝒛�, the mechanical momentum-
density acquired by the magnetized cylinder will be −𝜀𝜀0 𝑀𝑀0 𝐸𝐸𝑥𝑥 𝒚𝒚 �, which is the same mechanical
momentum as obtained in the Lorentz formalism — albeit after invoking the notion of hidden
momentum, an endemic feature of the Lorentz formulation in its treatment of magnetic materials.

4. Rotating magnetic dipole in a constant and uniform electric field. As our third example,
we consider a small magnetic dipole 𝒎𝒎0 residing at the origin of the 𝑥𝑥𝑥𝑥-plane and rotating
slowly around the 𝑧𝑧-axis at the constant angular velocity 𝜔𝜔, as shown in Fig. 3. We have
𝒎𝒎(𝑡𝑡) = 𝑚𝑚0 [cos(𝜔𝜔𝜔𝜔) 𝒙𝒙
� + sin(𝜔𝜔𝜔𝜔) 𝒚𝒚
�]. (46)
The dipole in Fig.3 sits between a pair of non-conducting, uniformly-charged, infinitely-
large parallel plates, which produce a constant and uniform electric field 𝑬𝑬0 = (𝜎𝜎0 ⁄𝜀𝜀0 )𝒙𝒙
� in the
region of space between the two plates. The distance between the plates is 𝑑𝑑, and the surface
electrical charge-densities of the two plates are assumed to be ±𝜎𝜎0 . The slow rotation of the
dipole ensures that the EM radiation is negligible, and that quasi-static treatment of the EM fields
is permitted. (Of course, for a magnetic dipole to rotate at a constant angular velocity, a torque
must be applied to the dipole. We may assume that a constant, uniform magnetic field 𝐻𝐻0 𝒛𝒛� is

10
present at and around the origin of coordinates in the system of Fig.3. The presence of this
magnetic field will in no way affect the ensuing analysis.)
In the Einstein-Laub formulation [33], the 𝐸𝐸-field exerts no torque on the dipole. However,
the dipole experiences a net force 𝑭𝑭(𝑡𝑡) in the presence of the external 𝐸𝐸-field. Denoting the
magnetization distribution within the dipole by 𝑴𝑴(𝒓𝒓, 𝑡𝑡), we will have
∞
𝑭𝑭(𝑡𝑡) = − ∭−∞ 𝜕𝜕𝑡𝑡 𝑴𝑴(𝒓𝒓, 𝑡𝑡) × 𝜀𝜀0 𝑬𝑬0 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 = 𝜎𝜎0 𝑚𝑚0 𝜔𝜔 cos(𝜔𝜔𝜔𝜔) 𝒛𝒛�. (47)
Next, we compute the EM (Abraham) momentum 𝒑𝒑𝑒𝑒𝑒𝑒 (𝑡𝑡) of the field [31-33], and show that
the force 𝑭𝑭(𝑡𝑡) is not rooted in an exchange of momentum between the EM field and the dipole.
z y
σ0
E0
+
−
+ −
+ + ωt − −
x
+ + −
m0 −
−
+ −
−σ 0

d
Fig. 3. A small magnetic dipole 𝒎𝒎0 rotates slowly in a constant, uniform 𝐸𝐸-field produced by a pair of
uniformly-charged parallel plates. The surface electric charge densities of the plates are denoted by ±𝜎𝜎0 .

Since the magnetic particle is assumed to rotate slowly, we may take the instantaneous
magnetization profile 𝑴𝑴(𝒓𝒓, 𝑡𝑡) and compute the corresponding magnetic charge-density
distribution 𝜌𝜌𝑚𝑚 (𝒓𝒓, 𝑡𝑡) = −𝜵𝜵 ∙ 𝑴𝑴(𝒓𝒓, 𝑡𝑡). The 𝐻𝐻-field distribution will then be obtained using
Maxwell’s magnetostatic equations [4-8], as follows:
∞
𝜌𝜌𝑚𝑚 (𝒓𝒓′ ,𝑡𝑡)(𝒓𝒓 − 𝒓𝒓′ )
𝑯𝑯(𝒓𝒓, 𝑡𝑡) = � 𝑑𝑑𝑥𝑥 ′ 𝑑𝑑𝑦𝑦 ′ 𝑑𝑑𝑧𝑧 ′ . (48)
4𝜋𝜋𝜇𝜇0 |𝒓𝒓 − 𝒓𝒓′ |3
−∞

In Eq.(48) 𝜇𝜇0 is the vacuum permeability. In the region of space between parallel plates of
Fig.3, the Abraham momentum of the EM field may be computed as follows:
+𝑑𝑑 ⁄2
1 ∞ ∞
𝒑𝒑𝑒𝑒𝑒𝑒 (𝑡𝑡) = 𝑐𝑐 2 � � ∫𝑧𝑧=−∞ 𝑬𝑬(𝒓𝒓, 𝑡𝑡) × 𝑯𝑯(𝒓𝒓, 𝑡𝑡)𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
𝑦𝑦=−∞
𝑥𝑥=−𝑑𝑑 ⁄2
∞
+𝑑𝑑 ⁄2
𝜎𝜎0 ∞ ∞ � × (𝒓𝒓 − 𝒓𝒓′)
𝒙𝒙
= � 𝜌𝜌𝑚𝑚 (𝒓𝒓′ , 𝑡𝑡) �� ∫ ∫
𝑦𝑦=−∞ 𝑧𝑧=−∞ |𝒓𝒓 − 𝒓𝒓′ |3
𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑� 𝑑𝑑𝑥𝑥 ′ 𝑑𝑑𝑦𝑦 ′ 𝑑𝑑𝑧𝑧 ′
4𝜋𝜋
𝑥𝑥=−𝑑𝑑 ⁄2
−∞

11
 ∞
+𝑑𝑑 ⁄2
𝜎𝜎0 ′ ∞ ∞ (𝑦𝑦 − 𝑦𝑦′ )𝒛𝒛� − (𝑧𝑧 − 𝑧𝑧 ′ )𝒚𝒚
�
= 4𝜋𝜋
� 𝜌𝜌𝑚𝑚 (𝒓𝒓 , 𝑡𝑡) �� ∫ ∫
𝑦𝑦=−∞ 𝑧𝑧=−∞ [(𝑥𝑥−𝑥𝑥 ′ )2 +(𝑦𝑦−𝑦𝑦 ′ )2 +(𝑧𝑧−𝑧𝑧 ′ )2 ]3⁄2
𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑� 𝑑𝑑𝑥𝑥 ′ 𝑑𝑑𝑦𝑦 ′ 𝑑𝑑𝑧𝑧 ′
𝑥𝑥=−𝑑𝑑 ⁄2
−∞
+𝑑𝑑 ⁄2
∞ ∞ 0 ∞
𝜎𝜎0 ′ �
𝒚𝒚
= � 𝜌𝜌𝑚𝑚 (𝒓𝒓 , 𝑡𝑡) � � � � ′ 2 ′ 2 ′ 2
� � 𝑑𝑑𝑑𝑑
4𝜋𝜋 �(𝑥𝑥−𝑥𝑥 ) +(𝑦𝑦−𝑦𝑦 ) +(𝑧𝑧−𝑧𝑧 ) 𝑧𝑧=−∞
−∞ 𝑦𝑦=−∞
𝑥𝑥=−𝑑𝑑 ⁄2
∞
0 ∞
𝒛𝒛�
−� � � � 𝑑𝑑𝑑𝑑� 𝑑𝑑𝑑𝑑𝑑𝑑𝑥𝑥 ′ 𝑑𝑑𝑦𝑦 ′ 𝑑𝑑𝑧𝑧 ′ = 0. (49)
�(𝑥𝑥−𝑥𝑥 ′ )2 +(𝑦𝑦−𝑦𝑦 ′ )2 +(𝑧𝑧−𝑧𝑧 ′ ) 𝑦𝑦=−∞
2
𝑧𝑧=−∞

In the absence of a net EM momentum in the region between the plates, the force 𝑭𝑭(𝑡𝑡) of
Eq.(47) cannot be balanced by a corresponding change in the field momentum 𝒑𝒑𝑒𝑒𝑒𝑒 (𝑡𝑡). We must,
therefore, look elsewhere for confirmation of momentum conservation. Specifically, we note that
the rotating dipole’s magnetic field, being time-dependent, must give rise to an 𝐸𝐸-field which
would then exert a force on the electrically-charged parallel plates. Since, in the absence of
retardation effects, the vector potential 𝑨𝑨(𝒓𝒓, 𝑡𝑡) of the slowly-rotating magnetic dipole is given by
𝑚𝑚
𝑨𝑨(𝒓𝒓, 𝑡𝑡) = 4𝜋𝜋𝑟𝑟03 [cos(𝜔𝜔𝜔𝜔) 𝒙𝒙
� + sin(𝜔𝜔𝜔𝜔) 𝒚𝒚
�] × 𝒓𝒓
𝑚𝑚
= 4𝜋𝜋𝑟𝑟03 [cos(𝜔𝜔𝜔𝜔) (𝑦𝑦𝒛𝒛� − 𝑧𝑧𝒚𝒚
�) + sin(𝜔𝜔𝜔𝜔) (𝑧𝑧𝒙𝒙
� − 𝑥𝑥𝒛𝒛�)], (50)
the corresponding 𝐸𝐸-field is readily evaluated as follows:
0 𝑚𝑚 𝜔𝜔
𝑬𝑬(𝒓𝒓, 𝑡𝑡) = −𝜕𝜕𝑡𝑡 𝑨𝑨(𝒓𝒓, 𝑡𝑡) = 4𝜋𝜋𝑟𝑟 3
[sin(𝜔𝜔𝜔𝜔) (𝑦𝑦𝒛𝒛� − 𝑧𝑧𝒚𝒚
�) − cos(𝜔𝜔𝜔𝜔) (𝑧𝑧𝒙𝒙
� − 𝑥𝑥𝒛𝒛�)]. (51)

To determine the force exerted by the above 𝐸𝐸-field on each of the charged plates, we fix the
coordinate 𝑥𝑥 and integrate the field over the 𝑦𝑦𝑦𝑦-plane to obtain
∞
∬−∞ 𝑬𝑬(𝒓𝒓, 𝑡𝑡)𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
∞
𝑚𝑚0 𝜔𝜔 𝑧𝑧 𝑥𝑥 𝑦𝑦
=− 4𝜋𝜋
� �𝑟𝑟 3 [cos(𝜔𝜔𝜔𝜔) 𝒙𝒙
� + sin(𝜔𝜔𝜔𝜔)𝒚𝒚
�] − � 3 cos(𝜔𝜔𝜔𝜔) + 3 sin(𝜔𝜔𝜔𝜔)� 𝒛𝒛�� 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
𝑟𝑟 𝑟𝑟
−∞
∞ ∞ 0
𝑚𝑚0 𝜔𝜔 1
= [cos(𝜔𝜔𝜔𝜔) 𝒙𝒙
� + sin(𝜔𝜔𝜔𝜔)𝒚𝒚
�] � � � � 𝑑𝑑𝑑𝑑
4𝜋𝜋 �𝑥𝑥 2 +𝑦𝑦 2 +𝑧𝑧 2
𝑦𝑦=−∞ 𝑧𝑧=−∞
∞
∞ 0 ∞
𝑚𝑚0 𝜔𝜔 1 𝑚𝑚0 𝜔𝜔 1
− sin(𝜔𝜔𝜔𝜔)𝒛𝒛� � � � � 𝑑𝑑𝑑𝑑 + 𝑥𝑥 cos(𝜔𝜔𝜔𝜔) 𝒛𝒛� � 𝑟𝑟 3
𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
4𝜋𝜋 �𝑥𝑥 2 +𝑦𝑦2 +𝑧𝑧 2 4𝜋𝜋
𝑦𝑦=−∞ −∞
𝑧𝑧=−∞
∞ ∞
𝑚𝑚0 𝜔𝜔 2𝜋𝜋𝜋𝜋𝜋𝜋𝜋𝜋 1
= 𝑥𝑥 cos(𝜔𝜔𝜔𝜔) 𝒛𝒛� � (𝑥𝑥 2 +𝜌𝜌2 )3⁄2
= −½𝑚𝑚0 𝜔𝜔𝜔𝜔 cos(𝜔𝜔𝜔𝜔) 𝒛𝒛� � ��
4𝜋𝜋 �𝑥𝑥 2 +𝜌𝜌 2
0 𝜌𝜌=0

= ½𝑚𝑚0 𝜔𝜔 cos(𝜔𝜔𝜔𝜔)(𝑥𝑥⁄|𝑥𝑥|) 𝒛𝒛�. (52)

For the plate on the left-hand-side in Fig.3, 𝑥𝑥 is negative and 𝜎𝜎0 is positive, whereas for the
plate on the right-hand-side, 𝑥𝑥 is positive while 𝜎𝜎0 is negative. The total force on both plates is,

12
therefore, 𝑭𝑭(𝑡𝑡) = −𝜎𝜎0 𝑚𝑚0 𝜔𝜔 cos(𝜔𝜔𝜔𝜔) 𝒛𝒛�. This is exactly equal in magnitude and opposite in sign to
the force exerted on the magnetic dipole; see Eq.(47). Note that the force exerted by the rotating
magnetic dipole on the parallel plates is independent of the separation 𝑑𝑑 between the plates.
Consequently, the effect of the dipole on the plates cannot be ignored, even if the plates happen
to be infinitely far away. We have thus confirmed the conservation of momentum by showing
explicitly that the action on the dipole is equal and opposite to the reaction of the parallel plates.

4.1. Hidden momentum of the rotating magnetic dipole. Let us now consider the same
problem from the perspective of the Lorentz formulation [1]. In this case, since the dipole has no
electric charge, it does not experience any force from the 𝐸𝐸-field, nor does the bound-current of
the dipole experience any force in the absence of an external 𝐵𝐵-field. So, it might be thought
that, in the system of Fig.3, the Lorentz formulation predicts no forces acting on the dipole. Now,
in the Lorentz formulation, the EM momentum is the Livens momentum [34], whose density is
𝜀𝜀0 𝑬𝑬(𝒓𝒓, 𝑡𝑡) × 𝑩𝑩(𝒓𝒓, 𝑡𝑡). Since 𝑩𝑩 = 𝜇𝜇0 𝑯𝑯 + 𝑴𝑴, and, in accordance with Eq.(49), the 𝜇𝜇0 𝑯𝑯 part of 𝑩𝑩
does not contribute to the EM momentum, the Livens momentum in the system of Fig.3 must be
(Livens)
𝒑𝒑𝑒𝑒𝑒𝑒 = 𝜀𝜀0 𝑬𝑬0 × 𝒎𝒎(𝑡𝑡) = 𝜎𝜎0 𝑚𝑚0 sin(𝜔𝜔𝜔𝜔) 𝒛𝒛�. (53)
[The same result as in Eq.(53) may be obtained using an alternative method of calculation
outlined in Appendix B.] It is thus tempting to associate the time-rate-of-change of the Livens
momentum of Eq.(53) with the mechanical force exerted on the parallel plates, and to declare
that momentum conservation is upheld.
However, in the Lorentz formulation there is known to be a certain amount of hidden
momentum 𝒑𝒑hidden = 𝒎𝒎(𝑡𝑡) × 𝜀𝜀0 𝑬𝑬0 = −𝜎𝜎0 𝑚𝑚0 sin(𝜔𝜔𝜔𝜔) 𝒛𝒛� associated with the magnetic dipole
residing in the external 𝐸𝐸-field [29-32]. The time-rate-of-change of the hidden momentum being
given by 𝜕𝜕𝑡𝑡 𝒑𝒑hidden = −𝜎𝜎0 𝑚𝑚0 𝜔𝜔 cos(𝜔𝜔𝜔𝜔) 𝒛𝒛�, it is clear that an equal and opposite force must be
experienced by the rotating dipole. The EM force experienced by the dipole in the Lorentz
formulation, assuming the contribution of hidden momentum is properly taken into account, is
thus seen to be in agreement with the result obtained in Eq.(47) via the Einstein-Laub
formulation. Once again the total momentum is seen to be conserved, albeit with the inclusion of
the hidden momentum and the Livens momentum in addition to the mechanical momenta
imparted to the magnetic dipole and to the pair of parallel plates.

5. Rotating electric dipole in a constant and uniform magnetic field. In this section we
examine the dual of the system of Fig.3, namely, a rotating electric dipole in a uniform magnetic
field. With reference to Fig.4, consider a constant electric dipole 𝒑𝒑0 , located at the origin of the
coordinate system and rotating slowly at a fixed angular velocity 𝜔𝜔 in the 𝑥𝑥𝑥𝑥-plane around the 𝑦𝑦-
axis. The dipole moment may thus be written
𝒑𝒑(𝑡𝑡) = 𝑝𝑝0 [cos(𝜔𝜔𝜔𝜔) 𝒙𝒙
� + sin(𝜔𝜔𝜔𝜔) 𝒛𝒛�]. (54)
In the presence of a constant, uniform magnetic field 𝑯𝑯(𝒓𝒓, 𝑡𝑡) = 𝐻𝐻0 𝒛𝒛�, produced by an
infinitely-long cylinder of radius 𝑅𝑅𝑐𝑐 carrying a constant surface-current-density 𝑱𝑱𝑠𝑠 = 𝐽𝐽0 𝝋𝝋
� = 𝐻𝐻0 𝝋𝝋
�,
the Lorentz force law yields the EM force exerted on the dipole as follows:
�.
𝑭𝑭(𝑡𝑡) = 𝜕𝜕𝑡𝑡 𝒑𝒑(𝑡𝑡) × 𝜇𝜇0 𝑯𝑯 = 𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 𝜔𝜔 sin(𝜔𝜔𝜔𝜔) 𝒚𝒚 (55)

13
 To confirm the conservation of momentum, we must first evaluate the EM counter-force
exerted by the magnetic field of the rotating dipole on the current-carrying cylinder. Considering
that the polarization density of the point-dipole is 𝑷𝑷(𝒓𝒓, 𝑡𝑡) = 𝒑𝒑(𝒕𝒕)𝛿𝛿(𝑥𝑥)𝛿𝛿(𝑦𝑦)𝛿𝛿(𝑧𝑧), its bound
current-density [4,5] is given by
𝑱𝑱bound (𝑡𝑡) = 𝜕𝜕𝑡𝑡 𝑷𝑷(𝒓𝒓, 𝑡𝑡) = −𝑝𝑝0 𝜔𝜔𝛿𝛿(𝑥𝑥)𝛿𝛿(𝑦𝑦)𝛿𝛿(𝑧𝑧)[sin(𝜔𝜔𝜔𝜔) 𝒙𝒙
� − cos(𝜔𝜔𝜔𝜔) 𝒛𝒛�]. (56)
Consequently, the vector potential 𝑨𝑨(𝒓𝒓, 𝑡𝑡) of the dipole in the quasi-static limit may be written as
𝜇𝜇0 𝑝𝑝0 𝜔𝜔
𝑨𝑨(𝒓𝒓, 𝑡𝑡) = − [sin(𝜔𝜔𝜔𝜔) 𝒙𝒙
� − cos(𝜔𝜔𝜔𝜔) 𝒛𝒛�]. (57)
4𝜋𝜋𝜋𝜋
The magnetic field produced in the surrounding region by the (slowly) rotating dipole is thus
given by
𝜇𝜇0 𝑝𝑝0 𝜔𝜔
𝑩𝑩(𝒓𝒓, 𝑡𝑡) = 𝜇𝜇0 𝑯𝑯(𝒓𝒓, 𝑡𝑡) = 𝜵𝜵 × 𝑨𝑨(𝒓𝒓, 𝑡𝑡) = [(𝑧𝑧𝒚𝒚
� − 𝑦𝑦𝒛𝒛�) sin(𝜔𝜔𝜔𝜔) − (𝑦𝑦𝒙𝒙
� − 𝑥𝑥𝒚𝒚
�) cos(𝜔𝜔𝜔𝜔)]. (58)
4𝜋𝜋𝑟𝑟 3

Fig. 4. A small spherical particle of radius 𝑅𝑅𝑠𝑠 centered at the origin of z

coordinates has a permanent electric dipole moment 𝒑𝒑0 , which rotates
slowly at a constant angular velocity 𝜔𝜔 in the 𝑥𝑥𝑥𝑥-plane around the 𝑦𝑦- H0 H0
axis. The dipole is acted upon by a constant, uniform magnetic field
𝐻𝐻0 𝒛𝒛�, produced by an infinitely long, thin cylinder of radius 𝑅𝑅𝑐𝑐 , which
carries the constant surface-current-density 𝑱𝑱𝑠𝑠 = 𝐽𝐽0 𝝋𝝋 � = 𝐻𝐻0 𝝋𝝋�. The
dipole experiences the oscillatory Lorentz force along the 𝑦𝑦-axis p0
given by Eq.(55).
ωt
We are now in a position to calculate the Lorentz y
force exerted by the rotating dipole on the cylinder of
radius 𝑅𝑅𝑐𝑐 carrying the surface-current-density 𝑱𝑱𝑠𝑠 = 𝐽𝐽0 𝝋𝝋
�= Js
� , which generates the magnetic field 𝐻𝐻0 𝒛𝒛� inside the
𝐻𝐻0 𝝋𝝋
cylinder. We have Rc
x
∞ 2𝜋𝜋
𝑭𝑭𝑐𝑐 (𝑡𝑡) = ∫𝑧𝑧=−∞ ∫𝜑𝜑=0 𝐽𝐽0 𝝋𝝋
� × 𝜇𝜇0 𝑯𝑯(𝑅𝑅𝑐𝑐 , 𝜑𝜑, 𝑧𝑧, 𝑡𝑡)𝑅𝑅𝑐𝑐 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
∞ 2𝜋𝜋 𝜇𝜇0 𝑝𝑝0 𝜔𝜔
= ∫𝑧𝑧=−∞ ∫𝜑𝜑=0 𝐻𝐻0 (cos 𝜑𝜑 𝒚𝒚
� −sin 𝜑𝜑 𝒙𝒙
�) × [(𝑧𝑧𝒚𝒚
�− 𝑦𝑦𝒛𝒛�) sin(𝜔𝜔𝜔𝜔) − (𝑦𝑦𝒙𝒙
� − 𝑥𝑥𝒚𝒚
�) cos(𝜔𝜔𝜔𝜔)]𝑅𝑅𝑐𝑐 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
4𝜋𝜋𝑟𝑟 3
0
𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 𝜔𝜔𝑅𝑅𝑐𝑐 ∞ 2𝜋𝜋 𝑅𝑅 sin 𝜑𝜑 cos 𝜑𝜑
=− �−
�[sin(𝜔𝜔𝜔𝜔)𝒙𝒙 cos(𝜔𝜔𝜔𝜔)𝒛𝒛�] ∫𝑧𝑧=−∞ ∫𝜑𝜑=0 (𝑅𝑅𝑐𝑐 2 + 𝑧𝑧 2 )3⁄2 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
4𝜋𝜋 𝑐𝑐
0
∞ 2𝜋𝜋 [𝑥𝑥cos(𝜔𝜔𝜔𝜔) + 𝑧𝑧 sin(𝜔𝜔𝜔𝜔)] sin 𝜑𝜑 ∞ 2𝜋𝜋 𝑦𝑦 sin 𝜑𝜑
+𝒛𝒛� ∫𝑧𝑧=−∞ ∫𝜑𝜑=0 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 � ∫𝑧𝑧=−∞ ∫𝜑𝜑=0 2 2 3⁄2 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑�
+ sin(𝜔𝜔𝜔𝜔) 𝒚𝒚
(𝑅𝑅𝑐𝑐2 + 𝑧𝑧 2 )3⁄2 (𝑅𝑅 + 𝑧𝑧 ) 𝑐𝑐

𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 𝜔𝜔𝑅𝑅𝑐𝑐 ∞ 2𝜋𝜋 𝑅𝑅 sin2 𝜑𝜑

=− �∫𝑧𝑧=−∞ ∫𝜑𝜑=0 (𝑅𝑅2𝑐𝑐+ 𝑧𝑧 2 )3⁄2 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑� sin(𝜔𝜔𝜔𝜔) 𝒚𝒚
�
4𝜋𝜋 𝑐𝑐

∞ 𝑑𝑑𝑑𝑑 ∞
� = −¼𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 𝜔𝜔�𝜁𝜁 ⁄�1 + 𝜁𝜁 2 ��−∞ sin(𝜔𝜔𝜔𝜔) 𝒚𝒚
= −¼𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 𝜔𝜔 �∫−∞ (1 + 𝜁𝜁2 )3⁄2 � sin(𝜔𝜔𝜔𝜔) 𝒚𝒚 �

�.
= −½𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 𝜔𝜔 sin(𝜔𝜔𝜔𝜔) 𝒚𝒚 (59)

The cylinder thus experiences a force only half as large as that experienced by the dipole,
and in the opposite direction; see Eq.(55). The missing momentum must, therefore, come from
the EM momentum inside the volume of the cylinder. To verify this, we take the dipole to be a

14
small spherical particle of radius 𝑅𝑅𝑠𝑠 and uniform polarization 𝑷𝑷(𝑡𝑡). In fact, it is best to imagine
two overlapping spherical dipoles, one oscillating along 𝑥𝑥, the other along the 𝑧𝑧-axis. Inside each
spherical dipole, the 𝐸𝐸-field is uniform and given by −𝑷𝑷(𝑡𝑡)/(3𝜀𝜀0 ). The EM momentum inside
the spherical dipole oscillating along the 𝑥𝑥-axis is thus equal to 𝓹𝓹𝑒𝑒𝑒𝑒 (𝑡𝑡) = ⅓𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 cos(𝜔𝜔𝑡𝑡) 𝒚𝒚
�,
whereas that inside the dipole oscillating along the 𝑧𝑧-axis is zero. The EM momentum outside
the spherical dipole oscillating along 𝑧𝑧 also turns out to be zero. To see this, note that the quasi-
static scalar potential 𝜓𝜓(𝒓𝒓, 𝑡𝑡) associated with 𝒑𝒑(𝑡𝑡) = 𝑝𝑝0 sin(𝜔𝜔𝜔𝜔) 𝒛𝒛� in the region outside the
sphere of radius 𝑅𝑅𝑠𝑠 is given by [4-8]
𝑝𝑝0 sin(𝜔𝜔𝜔𝜔)𝑧𝑧 𝑝𝑝 sin(𝜔𝜔𝜔𝜔)𝑧𝑧
𝜓𝜓(𝒓𝒓, 𝑡𝑡) = 4𝜋𝜋𝜀𝜀0 𝑟𝑟 3
= 4𝜋𝜋𝜀𝜀 0(𝜌𝜌2 + 𝑧𝑧 2 )3⁄2. (60)
0

In the above equation, 𝑟𝑟 is the distance from the origin to the observation point (spherical
coordinates), whereas 𝜌𝜌 is the radial distance from the 𝑧𝑧-axis (cylindrical coordinates). The
corresponding 𝐸𝐸-field and EM momentum-density inside the cylinder are thus found to be
� − (𝜕𝜕𝑧𝑧 𝜓𝜓)𝒛𝒛�.
𝑬𝑬(𝒓𝒓, 𝑡𝑡) = −𝜵𝜵𝜓𝜓(𝒓𝒓, 𝑡𝑡) = −(𝜕𝜕𝜌𝜌 𝜓𝜓)𝝆𝝆 (61)
3𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 sin(𝜔𝜔𝜔𝜔)𝑧𝑧𝑧𝑧
𝑬𝑬(𝒓𝒓, 𝑡𝑡) × 𝑯𝑯(𝒓𝒓, 𝑡𝑡)⁄𝑐𝑐 2 = − � 4𝜋𝜋(𝜌𝜌2 + 𝑧𝑧 2 )5⁄2
�.
� 𝝋𝝋 (62)

Clearly, the integral of the momentum-density in Eq.(62) over the cylinder’s volume
(excluding the particle’s spherical volume) vanishes. We conclude that the dipole oscillating
along the 𝑧𝑧-axis does not contribute to the EM momentum of the system of Fig.4. As for the
dipole oscillating along the 𝑥𝑥-axis, we have (in the region inside the cylinder and outside the
spherical particle)
𝑝𝑝0 cos(𝜔𝜔𝜔𝜔)𝑥𝑥 𝑝𝑝0 cos(𝜔𝜔𝜔𝜔)𝜌𝜌 cos 𝜑𝜑
𝜓𝜓(𝒓𝒓, 𝑡𝑡) = 4𝜋𝜋𝜀𝜀0 𝑟𝑟 3
= 4𝜋𝜋𝜀𝜀0 (𝜌𝜌2 + 𝑧𝑧 2 )3⁄2
. (63)

� − 𝜌𝜌−1 (𝜕𝜕𝜑𝜑 𝜓𝜓)𝝋𝝋

𝑬𝑬(𝒓𝒓, 𝑡𝑡) = −𝜵𝜵𝜓𝜓(𝒓𝒓, 𝑡𝑡) = −(𝜕𝜕𝜌𝜌 𝜓𝜓)𝝆𝝆 � − (𝜕𝜕𝑧𝑧 𝜓𝜓)𝒛𝒛�. (64)

𝑬𝑬(𝒓𝒓, 𝑡𝑡) × 𝑯𝑯(𝒓𝒓, 𝑡𝑡)⁄𝑐𝑐 2 = 𝜇𝜇0 𝜀𝜀0 𝐻𝐻0 ��𝜕𝜕𝜌𝜌 𝜓𝜓�𝝋𝝋

� − 𝜌𝜌−1 �𝜕𝜕𝜑𝜑 𝜓𝜓�𝝆𝝆
��
𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 cos(𝜔𝜔𝜔𝜔) 1 3𝜌𝜌2 sin 𝜑𝜑
= � + 2 2 3⁄2 𝝆𝝆
��(𝜌𝜌2 + 𝑧𝑧 2 )3⁄2 − (𝜌𝜌2 + 𝑧𝑧 2 )5⁄2 � cos 𝜑𝜑 𝝋𝝋 ��
4𝜋𝜋 (𝜌𝜌 + 𝑧𝑧 )

𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 cos(𝜔𝜔𝜔𝜔) 1 3𝜌𝜌2 sin 𝜑𝜑

= �� − � cos 𝜑𝜑 (cos 𝜑𝜑 𝒚𝒚
� − sin 𝜑𝜑 𝒙𝒙
�) + (cos 𝜑𝜑 𝒙𝒙
� + sin 𝜑𝜑 𝒚𝒚
�)�
4𝜋𝜋 (𝜌𝜌2 + 𝑧𝑧 2 )3⁄2 (𝜌𝜌2 + 𝑧𝑧 2 )5⁄2 (𝜌𝜌2 + 𝑧𝑧 2 )3⁄2

𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 cos(𝜔𝜔𝜔𝜔) �

𝒚𝒚 3𝜌𝜌2 �sin 𝜑𝜑 cos 𝜑𝜑 𝒙𝒙
� − cos2 𝜑𝜑 𝒚𝒚
��
= �(𝜌𝜌2 + 𝑧𝑧 2 )3⁄2 + (𝜌𝜌2 + 𝑧𝑧 2 )5⁄2
�. (65)
4𝜋𝜋

It is now straightforward to integrate the EM momentum-density of Eq.(65) over the volume

of the cylinder (excluding the interior of the spherical particle). The integral must be evaluated in
two steps, first when 𝜌𝜌 ranges from 𝑅𝑅𝑠𝑠 to 𝑅𝑅𝑐𝑐 , and then when 𝜌𝜌 goes from 0 to 𝑅𝑅𝑠𝑠 . We will have
𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 cos(𝜔𝜔𝜔𝜔) 𝑅𝑅𝑐𝑐 2𝜋𝜋 ∞ �
𝒚𝒚 3𝜌𝜌2 �sin 𝜑𝜑 cos 𝜑𝜑 �
𝒙𝒙 − cos2 𝜑𝜑 𝒚𝒚
��
a) 4𝜋𝜋
∫𝜌𝜌=𝑅𝑅 ∫𝜑𝜑=0 ∫𝑧𝑧=−∞ �(𝜌𝜌2 + 𝑧𝑧 2 )3⁄2 + (𝜌𝜌2 + 𝑧𝑧 2 )5⁄2
� 𝜌𝜌𝜌𝜌𝜌𝜌𝜌𝜌𝜌𝜌𝜌𝜌𝜌𝜌
𝑠𝑠

𝑐𝑐 𝑅𝑅 ∞ 𝜌𝜌 3𝜌𝜌3
� ∫𝜌𝜌=𝑅𝑅
= 𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 cos(𝜔𝜔𝜔𝜔) 𝒚𝒚 ∫𝑧𝑧=0 �(𝜌𝜌2 + 𝑧𝑧 2 )3⁄2 − 2(𝜌𝜌2 + 𝑧𝑧 2 )5⁄2 � 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 See Appendix A
𝑠𝑠

15
 𝑐𝑐 𝑅𝑅 1 1
� ∫𝜌𝜌=𝑅𝑅
= 𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 cos(𝜔𝜔𝜔𝜔) 𝒚𝒚 �𝜌𝜌 − 𝜌𝜌� 𝑑𝑑𝑑𝑑 = 0. (66a)
𝑠𝑠

𝑠𝑠 𝑅𝑅 ∞ 𝜌𝜌 3𝜌𝜌3
� ∫𝜌𝜌=0
b) 𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 cos(𝜔𝜔𝜔𝜔) 𝒚𝒚 ∫𝑧𝑧=�𝑅𝑅2 −𝜌𝜌2 �½ �(𝜌𝜌2 + 𝑧𝑧 2 )3⁄2 − 2(𝜌𝜌2 + 𝑧𝑧 2 )5⁄2 � 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 See Appendix A
𝑠𝑠
𝑅𝑅𝑠𝑠
1⁄2 1⁄2 3⁄2
1 �𝑅𝑅𝑠𝑠2 −𝜌𝜌2 � 3 �𝑅𝑅𝑠𝑠2 −𝜌𝜌2 � 1 �𝑅𝑅𝑠𝑠2 −𝜌𝜌2 �
� � � �1 −
= 𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 cos(𝜔𝜔𝜔𝜔) 𝒚𝒚 � − 2𝜌𝜌 �1 − −3+ �� 𝑑𝑑𝑑𝑑
𝜌𝜌 𝑅𝑅𝑠𝑠 𝑅𝑅𝑠𝑠 3𝑅𝑅𝑠𝑠3
0
𝑅𝑅𝑠𝑠
� � �𝜌𝜌�𝑅𝑅𝑠𝑠2 − 𝜌𝜌2 �𝑅𝑅𝑠𝑠3 �𝑑𝑑𝑑𝑑
= ½𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 cos(𝜔𝜔𝜔𝜔) 𝒚𝒚
0
1 1
� � 𝜂𝜂�1 − 𝜂𝜂2 𝑑𝑑𝑑𝑑 = 𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 cos(𝜔𝜔𝜔𝜔) 𝒚𝒚
= ½𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 cos(𝜔𝜔𝜔𝜔) 𝒚𝒚 �. See Appendix A (66b)
0 6

The total EM momentum (including the contribution from the interior region of the spherical
dipole) is thus seen to be 𝓹𝓹𝑒𝑒𝑒𝑒 (𝑡𝑡) = ½𝜇𝜇0 𝐻𝐻0 𝑝𝑝0 cos(𝜔𝜔𝜔𝜔) 𝒚𝒚
�, confirming the conservation of overall
momentum (i.e., electromagnetic plus mechanical) in the system of Fig.4.

6. Energy of a dipole in an external EM field. Our next example concerns the EM energy of an
electric dipole in a constant and uniform external electric field. The E-field of a small,
uniformly-polarized spherical particle (i.e., spherical dipole) located at the origin of a spherical
coordinate system and oriented along the z-axis is given by [4-8]
�)
𝑝𝑝0 (2 cos 𝜃𝜃𝒓𝒓� + sin 𝜃𝜃𝜽𝜽
4𝜋𝜋𝜀𝜀0 𝑟𝑟 3
, 𝑟𝑟 ≥ 𝑅𝑅
𝑬𝑬(𝒓𝒓) = � (67)
𝑝𝑝 𝒛𝒛�
− 4𝜋𝜋𝜀𝜀0 𝑅𝑅3 , 𝑟𝑟 < 𝑅𝑅.
0

Here 𝑅𝑅 is the radius of the spherical particle and 𝑝𝑝0 𝒛𝒛� is its (electric) dipole moment. In a
static external electric field 𝑬𝑬(ext) (𝒓𝒓), the energy of the dipole is said to be
(ext)
ℰ = −𝒑𝒑 ∙ 𝑬𝑬(ext) (0) = −𝑝𝑝0 𝐸𝐸𝑧𝑧 (0). (68)
This is because the torque exerted by 𝑬𝑬(ext) (𝒓𝒓) on the dipole 𝒑𝒑 oriented in an arbitrary
direction is given by 𝝉𝝉 = 𝒑𝒑 × 𝑬𝑬(ext) (0). Denoting the angle between 𝒑𝒑 and 𝑬𝑬(ext) (0) by 𝜑𝜑, the
magnitude of the torque is readily seen to be 𝑝𝑝0 𝐸𝐸 (ext) (0) sin 𝜑𝜑. If now the angle 𝜑𝜑 changes by
𝛿𝛿𝛿𝛿, the work done by the E-field on the dipole will be
𝛿𝛿𝛿𝛿 = 𝑝𝑝0 𝐸𝐸 (ext) (0) sin 𝜑𝜑 𝛿𝛿𝛿𝛿 = −𝛿𝛿�𝑝𝑝0 𝐸𝐸 (ext) (0) cos 𝜑𝜑� = −𝛿𝛿�𝒑𝒑 ∙ 𝑬𝑬(ext) (0)�. (69)
The energy of the dipole in the external E-field is thus said to be −𝒑𝒑 ∙ 𝑬𝑬(ext) (0), with the
zero of energy assigned (arbitrarily) to that orientation of the dipole which is perpendicular to
𝑬𝑬(ext) (0). This energy is exchanged between the dipole’s rotational kinetic energy and the stored
energy in the E-field in the region of space surrounding the dipole. To see this, note that the
energy-density of the E-field is generally given by [4,5]
2
ℰ(𝒓𝒓) = ½𝜀𝜀0 |𝑬𝑬total (𝒓𝒓)|2 = ½𝜀𝜀0 �𝑬𝑬(𝒓𝒓) + 𝑬𝑬(ext) (𝒓𝒓)�
2
= ½𝜀𝜀0 |𝑬𝑬(𝒓𝒓)|2 + ½𝜀𝜀0 �𝑬𝑬(ext) (𝒓𝒓)� + 𝜀𝜀0 𝑬𝑬(𝒓𝒓) ∙ 𝑬𝑬(ext) (𝒓𝒓). (70)

16
 The first two terms on the right-hand-side of the above equation are independent of the
orientation of the dipole. Thus the stored energy in the E-field varies as the integral of the third
term, 𝜀𝜀0 𝑬𝑬(𝒓𝒓) ∙ 𝑬𝑬(ext) (𝒓𝒓), over the entire space.
When the dipole is aligned with the z-axis, its E-field given by Eq.(67) maybe equivalently
expressed as follows:
𝑝𝑝 𝜕𝜕 𝒓𝒓
𝑬𝑬(𝒓𝒓) = − 4𝜋𝜋𝜀𝜀0 � �.
𝜕𝜕𝜕𝜕 𝑟𝑟 3
(71)
0

This should be obvious, considering that the dipole consists of a pair of point-charges, ±𝑞𝑞,
separated by a short distance 𝑑𝑑 along the z-axis. The E-field produced by a charge 𝑞𝑞 located at
the origin of the coordinate system is given by 𝑞𝑞𝒓𝒓/(4𝜋𝜋𝜀𝜀0 𝑟𝑟 3 ). Separating the ±𝑞𝑞 charges by a
short distance 𝑑𝑑 along the z-axis and recalling that 𝑝𝑝0 = 𝑞𝑞𝑞𝑞 yields the dipolar E-field of Eq.(71).
Note, however, that Eq.(71) does not exactly reproduce the internal field of the dipole in the
region 𝑟𝑟 < 𝑅𝑅 as specified in Eq.(67). Nevertheless, in the vicinity of the origin, the general
behavior of the E-field of Eq.(71) is consistent with that of the spherical dipole described by
Eq.(67).
The relevant part of the E-field energy in Eq.(70), the third term on the right-hand-side (i.e.,
the cross-term), may now be written as follows:
∞ 𝑝𝑝 ∞ 𝜕𝜕 𝒓𝒓
∆ℰ = ∭−∞ 𝜀𝜀0 𝑬𝑬(𝒓𝒓) ∙ 𝑬𝑬(ext) (𝒓𝒓)𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 = − 4𝜋𝜋0 ∭−∞ 𝜕𝜕𝜕𝜕 �𝑟𝑟 3 � ∙ 𝑬𝑬(ext) (𝒓𝒓)𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
𝑝𝑝 ∞ 𝒓𝒓 𝜕𝜕𝑬𝑬(ext) (𝒓𝒓)
= 4𝜋𝜋0 ∭−∞ 𝑟𝑟 3 ∙ 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑. (72)
𝜕𝜕𝜕𝜕
The final step in the above derivation involves an integration by parts followed by the
assumption that the external E-field at ±∞ is sufficiently weak to be negligible. We now invoke
(ext) (ext)
the static nature of 𝑬𝑬(ext) (𝒓𝒓) and the fact that 𝜵𝜵 × 𝑬𝑬(ext) (𝒓𝒓) = 0 to replace 𝜕𝜕𝑧𝑧 𝐸𝐸𝑦𝑦 by 𝜕𝜕𝑦𝑦 𝐸𝐸𝑧𝑧
(ext) (ext)
and 𝜕𝜕𝑧𝑧 𝐸𝐸𝑥𝑥 by 𝜕𝜕𝑥𝑥 𝐸𝐸𝑧𝑧 . We will have
𝑝𝑝 ∞ 𝒓𝒓 (ext) (ext) (ext)
∆ℰ = 4𝜋𝜋0 ∭−∞ 𝑟𝑟 3 ∙ �𝜕𝜕𝑥𝑥 𝐸𝐸𝑧𝑧 � + 𝜕𝜕𝑦𝑦 𝐸𝐸𝑧𝑧
𝒙𝒙 � + 𝜕𝜕𝑧𝑧 𝐸𝐸𝑧𝑧
𝒚𝒚 𝒛𝒛��𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
𝑝𝑝 ∞ 𝒓𝒓 (ext)
= − 4𝜋𝜋0 ∭−∞ �𝜵𝜵 ∙ �𝑟𝑟 3 �� 𝐸𝐸𝑧𝑧 (𝒓𝒓)𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
𝑝𝑝 ∞ (ext) (ext)
= − 4𝜋𝜋0 ∭−∞ 4𝜋𝜋𝜋𝜋(𝒓𝒓)𝐸𝐸𝑧𝑧 (𝒓𝒓)𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 = −𝑝𝑝0 𝐸𝐸𝑧𝑧 (0). (73)

z −−−−−−−−−−−−−−−−−
Fig.5. A small spherical dipole 𝒑𝒑 = 𝑝𝑝0 𝒛𝒛� is placed
in the uniform E-field between two uniformly- E0 p
charged parallel plates. The EM energy stored in
the E-field contains a contribution from the cross-
x y ++++++++++++++++++
term 𝜀𝜀0 𝑬𝑬(𝒓𝒓) ∙ 𝑬𝑬(ext) (𝒓𝒓), where 𝑬𝑬(ext) (𝒓𝒓) = 𝐸𝐸0 𝒛𝒛�.

Thus, when the dipole is anti-parallel to 𝑬𝑬(ext) (0), the energy stored in the E-field is at its
peak value. Allowing the dipole to rotate will enable the torque to bring it into alignment with
the local external field. In the process the dipole acquires rotational kinetic energy while the
energy stored in the E-field is reduced by an equal amount. A simple scenario involving a dipole
𝒑𝒑 placed between two uniformly-charged, infinitely large, parallel plates, as shown in Fig.5,

17
would be instructive. It is easy to verify by direct integration that, when 𝒑𝒑 = 𝑝𝑝0 𝒛𝒛� and the dipolar
E-field 𝑬𝑬(𝒓𝒓) is given by Eq.(67) while 𝑬𝑬(ext) (𝒓𝒓) = 𝐸𝐸0 𝒛𝒛� in the region between the parallel plates,
the cross-term 𝜀𝜀0 𝑬𝑬(𝒓𝒓) ∙ 𝑬𝑬(ext) (𝒓𝒓) appearing in Eq.(70) would integrate to −𝑝𝑝0 𝐸𝐸0 .

6.1. The case of a relativistically-induced electric dipole. Next, with reference to Fig.6,
consider a small, uniformly-magnetized, spherical particle (i.e., a magnetic dipole) located at the
origin of the 𝑥𝑥′𝑦𝑦′𝑧𝑧′ coordinate system, with its magnetization aligned with the 𝑧𝑧 ′ -axis. The
magnetic field of the dipole is given by
�)
𝑚𝑚0 (2 cos 𝜃𝜃 ′ 𝒓𝒓� + sin 𝜃𝜃′ 𝜽𝜽
3 , 𝑟𝑟 ′ ≥ 𝑅𝑅
′) 4𝜋𝜋𝜇𝜇0 𝑟𝑟 ′
𝑯𝑯(𝒓𝒓 = � 𝑚𝑚0 𝒛𝒛�
(74)
− 4𝜋𝜋𝜇𝜇 , 𝑟𝑟 ′ < 𝑅𝑅.
0 𝑅𝑅 3

z z′

Fig.6. A small spherical magnetic dipole 𝒎𝒎 = 𝑚𝑚0 𝒛𝒛� m V

moves with constant velocity 𝑉𝑉 along the y-axis of a
y′
Cartesian coordinate system. The rest-frame of the
y
particle is identified as 𝑥𝑥 ′ 𝑦𝑦 ′ 𝑧𝑧 ′ . x′
x
Note that, in order to maintain complete symmetry with the case of an electric dipole, we are
defining the magnetic dipole moment 𝑚𝑚0 𝒛𝒛� and its corresponding magnetization 𝑴𝑴 =
3𝑚𝑚0 𝒛𝒛�/(4𝜋𝜋𝑅𝑅 3 ) such that, in the SI system of units, 𝑩𝑩 = 𝜇𝜇0 𝑯𝑯 + 𝑴𝑴. In this way, the magnetic
dipole moment of a small loop of area 𝐴𝐴 carrying a constant electric current 𝐼𝐼 in the 𝑥𝑥𝑥𝑥-plane
will be 𝒎𝒎0 = 𝜇𝜇0 𝐼𝐼𝐼𝐼 𝒛𝒛�.
Suppose now that the 𝑥𝑥′𝑦𝑦′𝑧𝑧′ frame moves with constant velocity 𝑉𝑉 along the y-axis. From
the perspective of an observer in the stationary frame 𝑥𝑥𝑥𝑥𝑥𝑥 (i.e., the lab frame), the magnetic
�. The
dipole 𝑚𝑚0 𝒛𝒛� will be accompanied by a relativistically-induced electric dipole 𝒑𝒑 = 𝜀𝜀0 𝑉𝑉𝑚𝑚0 𝒙𝒙
E-field in the 𝑥𝑥𝑥𝑥𝑥𝑥 frame is readily obtained from a Lorentz transformation of the B-field in the
𝑥𝑥′𝑦𝑦′𝑧𝑧′ frame, as follows:
𝛾𝛾𝛾𝛾𝑚𝑚0
� + 𝛾𝛾𝛾𝛾𝐵𝐵𝑥𝑥 𝒛𝒛� =
−𝛾𝛾𝛾𝛾𝐵𝐵𝑧𝑧 𝒙𝒙 [(𝑥𝑥 2 + 𝛾𝛾 2 𝑦𝑦 2 − 2𝑧𝑧 2 )𝒙𝒙
� + 3𝑥𝑥𝑥𝑥𝒛𝒛�], 𝑟𝑟′ ≥ 𝑅𝑅
4𝜋𝜋(𝑥𝑥 2 +𝛾𝛾 2 𝑦𝑦 2 +𝑧𝑧 2 )5⁄2
𝑬𝑬(𝒓𝒓, 𝑡𝑡 = 0) = � (75)
𝛾𝛾𝛾𝛾𝑚𝑚0 ′
�=
−𝛾𝛾𝛾𝛾𝐵𝐵𝑧𝑧 𝒙𝒙 − �,
𝒙𝒙 𝑟𝑟 < 𝑅𝑅.
2𝜋𝜋𝑅𝑅3

Clearly, this field is very different from the E-field produced by an ordinary electric dipole
� placed at the origin of the 𝑥𝑥𝑥𝑥𝑥𝑥 coordinate system. In the presence of a static
𝒑𝒑 = 𝜀𝜀0 𝑉𝑉𝑚𝑚0 𝒙𝒙
external E-field 𝑬𝑬(ext) (𝒓𝒓), the E-field energy associated with the third term on the right-hand-
side of Eq.(70) will be
∞
∆ℰ(𝑡𝑡 = 0) = ∭−∞ 𝜀𝜀0 𝑬𝑬(𝑥𝑥, 𝑦𝑦, 𝑧𝑧, 𝑡𝑡 = 0) ∙ 𝑬𝑬(ext) (𝑥𝑥, 𝑦𝑦, 𝑧𝑧)𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
∞
= 𝜀𝜀0 𝛾𝛾 −1 ∭−∞ 𝑬𝑬(𝑥𝑥, 𝛾𝛾 −1 𝑦𝑦, 𝑧𝑧, 𝑡𝑡 = 0) ∙ 𝑬𝑬(ext) (𝑥𝑥, 𝛾𝛾 −1 𝑦𝑦, 𝑧𝑧)𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑. (76)
It is readily seen from Eq.(75) that, in the region 𝑟𝑟 ′ ≥ 𝑅𝑅, the dipolar E-field may be expressed as
𝛾𝛾𝛾𝛾𝑚𝑚0 𝜕𝜕 �−𝑥𝑥𝒛𝒛�
𝑧𝑧𝒙𝒙
𝑬𝑬(𝑥𝑥, 𝛾𝛾 −1 𝑦𝑦, 𝑧𝑧, 𝑡𝑡 = 0) = � 𝑟𝑟 3
�. (77)
4𝜋𝜋 𝜕𝜕𝜕𝜕

18
 The above expression, however, does not reproduce the correct behavior for the internal E-
field of the dipole (i.e., in the region 𝑟𝑟 ′ < 𝑅𝑅), as given by Eq.(75). The reason being that
𝜕𝜕𝑧𝑧 (𝒓𝒓⁄𝑟𝑟 3 ) is an appropriate representation for the H-field of the magnetic dipole, whereas the E-
field of the induced electric dipole given by Eq.(75) is intimately tied to the B-field of the
magnetic dipole. We thus proceed to evaluate ∆ℰ(𝑡𝑡 = 0) of Eq.(76) by ignoring the difference
between the E-field of Eq.(77) and the true internal field of the dipole. The result will
subsequently be adjusted to accommodate the correction to the internal field. Substituting
Eq.(77) into Eq.(76), we find
𝜀𝜀0 𝑉𝑉𝑚𝑚0 ∞ 𝜕𝜕 �−𝑥𝑥𝒛𝒛�
𝑧𝑧𝒙𝒙
∆ℰ(𝑡𝑡 = 0) = ∭−∞ 𝜕𝜕𝜕𝜕 � 𝑟𝑟 3
� ∙ 𝑬𝑬(ext) (𝑥𝑥, 𝛾𝛾 −1 𝑦𝑦, 𝑧𝑧)𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
4𝜋𝜋
(ext) (ext)
𝜀𝜀0 𝑉𝑉𝑚𝑚0 ∞ �−𝑥𝑥𝒛𝒛�
𝑧𝑧𝒙𝒙 𝜕𝜕𝐸𝐸𝑥𝑥 𝜕𝜕𝐸𝐸𝑧𝑧
=− ∭−∞ � �∙� �+
𝒙𝒙 𝒛𝒛�� 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
4𝜋𝜋 𝑟𝑟 3 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
(ext) (ext)
𝜀𝜀0 𝑉𝑉𝑚𝑚0 ∞ �−𝑥𝑥𝒛𝒛�
𝑧𝑧𝒙𝒙 𝜕𝜕𝐸𝐸𝑧𝑧 𝜕𝜕𝐸𝐸𝑧𝑧
=− ∭−∞ � �∙� �+
𝒙𝒙 𝒛𝒛�� 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
4𝜋𝜋 𝑟𝑟 3 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
𝜀𝜀0 𝑉𝑉𝑚𝑚0 ∞ 𝜕𝜕 𝑧𝑧 𝜕𝜕 𝑥𝑥 (ext)
= ∭−∞ �𝜕𝜕𝜕𝜕 �𝑟𝑟 3 � − 𝜕𝜕𝜕𝜕 �𝑟𝑟 3 �� 𝐸𝐸𝑧𝑧 (𝑥𝑥, 𝛾𝛾 −1 𝑦𝑦, 𝑧𝑧)𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 = 0. (78)
4𝜋𝜋

We must now add to the above result the correction due to the internal E-field of the induced
�. This correction arises from the difference between the internal B-
electric dipole 𝒑𝒑 = 𝜀𝜀0 𝑉𝑉𝑚𝑚0 𝒙𝒙
field of the magnetic dipole and its corresponding 𝜇𝜇0 𝐻𝐻-field, which is just the magnetization 𝑀𝑀
of the dipole. In the spherical dipole model, the correction amounts to an additional 𝑬𝑬 = −𝛾𝛾𝛾𝛾𝛾𝛾𝒙𝒙 �
′ (ext)
in the region 𝑟𝑟 < 𝑅𝑅, whose contribution to ∆ℰ(𝑡𝑡 = 0) is readily seen to be −𝒑𝒑 ∙ 𝑬𝑬 (0). In the
end, we recover the standard formula for the energy of an electric dipole immersed in a static
external E-field, even though the field distribution of a relativistically-induced electric dipole
turned out to be quite different from that of an ordinary electric dipole.

6.2. Magnetic dipole in an external magnetic field. It is instructive to examine the similarities
and differences between an electric dipole immersed in an external electric field, and a magnetic
dipole immersed in an external magnetic field. One major difference is that, while in the
preceding derivations involving a static E-field, we could use the identity
z
𝜵𝜵 × 𝑬𝑬(ext) (𝒓𝒓) = 0, the corresponding identity cannot be invoked for
magnetic fields, where 𝜵𝜵 × 𝑯𝑯(ext) (𝒓𝒓) = 𝑱𝑱free (𝒓𝒓). As a simple example, +
+ + +
consider a small, uniformly-magnetized spherical particle (i.e., a magnetic + +
+ +
dipole) placed inside a pair of infinitely-long, uniformly-charged − − − −
concentric cylinders whose radii are nearly equal and have equal but − −
opposite surface-charge-densities, as shown in Fig.7. −
−
Fig.7. A small spherical magnetic dipole 𝒎𝒎 = 𝑚𝑚0 𝒛𝒛� inside a pair of concentric, − −
infinitely-long, uniformly-charged cylinders. The electric charge-densities of the two − − − −
cylinders are equal in magnitude and opposite in sign. When the cylinders are spun in
opposite directions, a uniform magnetic field 𝑯𝑯(𝒓𝒓) = 𝐻𝐻0 𝒛𝒛� is created inside the cylinders.

The cylinders are initially at rest, but, at 𝑡𝑡 = 0, they begin to rotate slowly in opposite
directions. A uniform magnetic field 𝐻𝐻0 𝒛𝒛� thus builds up inside the cylinders, immersing the
magnetic dipole 𝑚𝑚0 𝒛𝒛� in an external magnetic field. Let us first analyze the system in accordance
with the Einstein-Laub formulation. Here no energy is exchanged between the dipole and the

19
cylinders as the latter spin up. Moreover, the H-field inside the cylinders, being the sum of the
dipole field and the uniform field set up by the rotating cylinders, does not make a net
contribution to the energy (via a cross-term) upon integrating the H-field energy-density over the
interior volume of the cylinders. In other words, given the dipolar field
�)
𝑚𝑚0 (2 cos 𝜃𝜃𝒓𝒓� + sin 𝜃𝜃𝜽𝜽
4𝜋𝜋𝜇𝜇0 𝑟𝑟 3
, 𝑟𝑟 ≥ 𝑅𝑅
𝑯𝑯(𝒓𝒓) = � (79)
𝑚𝑚 𝒛𝒛�
− 4𝜋𝜋𝜇𝜇0 𝑅𝑅3 , 𝑟𝑟 < 𝑅𝑅,
0

the total energy-density inside the cylinders will be

2
ℰ(𝒓𝒓) = ½𝜇𝜇0 |𝑯𝑯total (𝒓𝒓)|2 = ½𝜇𝜇0 �𝑯𝑯(ext) (𝒓𝒓) + 𝑯𝑯(𝒓𝒓)�

= ½𝜇𝜇0 𝐻𝐻02 + ½𝜇𝜇0 |𝑯𝑯(𝒓𝒓)|2 + 𝜇𝜇0 𝐻𝐻0 𝐻𝐻𝑧𝑧 (𝒓𝒓). (80)

Now, the third term on the right-hand-side of Eq.(80) integrates to zero over the volume of
the cylinder. An easy way to see this is to note that, for a point-dipole 𝑚𝑚0 𝒛𝒛�, the 𝐻𝐻-field is
𝑚𝑚 𝜕𝜕 𝒓𝒓
𝑯𝑯(𝒓𝒓) = − 4𝜋𝜋𝜇𝜇0 � �,
𝜕𝜕𝑧𝑧 𝑟𝑟 3
(81)
0

whose integral over 𝑧𝑧 from −∞ to +∞ ends up being zero. The question remains, however, as to
what happens to the dipole’s energy in the external field 𝐻𝐻0 𝒛𝒛�, because surely there will be a
torque acting on the dipole if it is turned away from the z-axis, and the action of this torque on
the rotating dipole should change its rotational kinetic energy in exactly the same way that an
external E-field exchanges energy with a rotating electric dipole.
To answer the above question, imagine the magnetic dipole losing its magnetization, say, as
a result of warming up slowly to beyond its Curie temperature. The changing magnetization
produces an E-field in the surrounding space in accordance with 𝜵𝜵 × 𝑬𝑬 = −𝜕𝜕𝑡𝑡 𝑩𝑩, which acts on
the rotating cylinders to speed them up, if the initial magnetization is pointing up — or to slow
them down, if the initial magnetization is downward. The kinetic energy thus imparted to the
cylinders can be shown to be exactly equal to ±𝑚𝑚0 𝐻𝐻0 . The energy of a magnetic dipole in an
external H-field is thus seen to be ℰ = −𝒎𝒎0 ∙ 𝑯𝑯(ext) (0), as expected from an argument based on
the torque exerted by the external H-field on the dipole. In contrast to the case of an external E-
field acting on an electric dipole, however, the energy associated with a magnetic dipole is not
exchanged with the surrounding H-field but, rather, is exchanged with the source of the external
H-field, which, in the present example, is the rotating pair of cylinders.
In the Lorentz formulation, the magnetic field energy-density is ℰ(𝒓𝒓, 𝑡𝑡) = ½𝜇𝜇0−1 |𝑩𝑩(𝒓𝒓, 𝑡𝑡)|2 . In
this case the stored energy-density in the cross-term, namely, 𝜇𝜇0−1 𝑩𝑩(ext) (𝒓𝒓) ∙ 𝑩𝑩(𝒓𝒓), integrates to
𝑚𝑚0 𝐻𝐻0 . However, the bound current-density associated with the magnetic dipole, i.e., 𝜇𝜇0−1 𝜵𝜵 × 𝑴𝑴,
now exchanges energy with the E-field produced by the cylinders as they spin up from rest to
their final rotational speed. (Note: In the Einstein-Laub formulation, the bound current of the
magnetic dipole does not participate in the exchange of energy with the fields.) The energy thus
extracted from the bound current during the spin-up process turns out to be exactly equal to
𝑚𝑚0 𝐻𝐻0 . Therefore, no net energy is given by the spinning cylinders to the dipole and its B-field
during spin-up. Once again, any energy exchanged in consequence of a rotation of the magnetic
dipole will be between the dipole and the spinning cylinders.

20
7. Concluding remarks. While the conservation laws of EM energy and linear and angular
momenta, being direct consequences of the energy-momentum stress tensor formulation(s) of
classical electrodynamics [1], are universal, their verification in specific situations may or may
not be straightforward. In this paper, we have examined a few non-trivial EM systems which are
sufficiently simple to be amenable to exact solutions of Maxwell’s equations in conjunction with
the electrodynamic laws of force, torque, energy, and momentum.
It is well known that a light pulse reflecting from a mirror or passing through an absorptive
or transparent object exchanges its EM energy and momentum with the material medium. The
source of the radiation does not participate in these exchanges, since the light pulse, once
detached from its source, is essentially autonomous. The same, however, cannot be said about
exchanges that take place between static (or quasi-static) fields and material media. As the
examples of the preceding sections have amply demonstrated, the (undetached) sources of the
EM fields in such situations could actively participate in the exchange of energy and momentum.
When magnetic materials interact with an external electric field, the standard (i.e., Lorentz)
formulation of classical electrodynamics could deviate from an alternative formulation of force,
torque, energy and momentum first proposed by A. Einstein and J. Laub in 1908 [33]. In such
cases we have analyzed the system under consideration from both perspectives, so that the reader
might appreciate the similarities and differences of the two formulations. There exist other
formulations of classical electrodynamics (e.g., Chu, Minkowski, Abraham), each built upon the
foundation of Maxwell’s macroscopic equations, but deviating from the standard formulation in
their structure of the corresponding energy-momentum stress tensor [34]. We have chosen to stay
away from these alternative formulations, as we believe them to be phenomenological rather than
rooted in realistic microscopic models of interaction between matter and EM fields.

Appendix A
Below we list (or evaluate) the various integrals and identities used throughout the paper.
𝐽𝐽𝑚𝑚−1 (𝑥𝑥) + 𝐽𝐽𝑚𝑚+1 (𝑥𝑥) = 2(𝑚𝑚/𝑥𝑥)𝐽𝐽𝑚𝑚 (𝑥𝑥) (Gradshteyn & Ryzhik* 8.471-1) (A1)
′ (𝑥𝑥)
𝐽𝐽𝑚𝑚−1 (𝑥𝑥) − 𝐽𝐽𝑚𝑚+1 (𝑥𝑥) = 2𝐽𝐽𝑚𝑚 (Gradshteyn & Ryzhik 8.471-2) (A2)
′ (𝑥𝑥)
(𝑚𝑚/𝑥𝑥) 𝐽𝐽𝑚𝑚 (𝑥𝑥) − 𝐽𝐽𝑚𝑚 = 𝐽𝐽𝑚𝑚+1 (𝑥𝑥) (Gradshteyn & Ryzhik 8.472-2) (A3)
2 (𝑥𝑥)𝑑𝑑𝑑𝑑
∫ 𝑥𝑥𝐽𝐽𝑚𝑚 2 (𝑥𝑥)
= ½𝑥𝑥 2 [𝐽𝐽𝑚𝑚 − 𝐽𝐽𝑚𝑚−1 (𝑥𝑥)𝐽𝐽𝑚𝑚+1 (𝑥𝑥)] (Gradshteyn & Ryzhik 5.54-2) (A4)

𝑥𝑥 𝑥𝑥
0 ′ (𝑥𝑥)𝑑𝑑𝑑𝑑
∫0 𝑥𝑥 2 𝐽𝐽𝑚𝑚 (𝑥𝑥)𝐽𝐽𝑚𝑚 = ½𝑥𝑥02 𝐽𝐽𝑚𝑚
2 (𝑥𝑥 ) 0 2 2
0 − ∫0 𝑥𝑥𝐽𝐽𝑚𝑚 (𝑥𝑥)𝑑𝑑𝑑𝑑 = ½𝑥𝑥0 𝐽𝐽𝑚𝑚−1 (𝑥𝑥0 )𝐽𝐽𝑚𝑚+1 (𝑥𝑥0 ). (A5)
𝑥𝑥 2
2 (𝑥𝑥)
∫0 𝑥𝑥�(𝑚𝑚/𝑥𝑥)2 𝐽𝐽𝑚𝑚 ′
0
+ 𝐽𝐽𝑚𝑚 (𝑥𝑥)�𝑑𝑑𝑑𝑑
𝑥𝑥 ′ (𝑥𝑥)]2 ′ (𝑥𝑥)
= ∫0 0 {𝑥𝑥[(𝑚𝑚/𝑥𝑥) 𝐽𝐽𝑚𝑚 (𝑥𝑥) − 𝐽𝐽𝑚𝑚 + 2𝑚𝑚𝐽𝐽𝑚𝑚 𝐽𝐽𝑚𝑚 (𝑥𝑥)}𝑑𝑑𝑑𝑑
𝑥𝑥 2 ′ (𝑥𝑥)
= ∫0 0 [𝑥𝑥 𝐽𝐽𝑚𝑚+1 (𝑥𝑥) + 2𝑚𝑚𝐽𝐽𝑚𝑚 𝐽𝐽𝑚𝑚 (𝑥𝑥)]𝑑𝑑𝑑𝑑
= ½𝑥𝑥02 { 𝐽𝐽𝑚𝑚+1
2 (𝑥𝑥0 ) − 𝐽𝐽𝑚𝑚 (𝑥𝑥0 )𝐽𝐽𝑚𝑚+2 (𝑥𝑥0 ) + 2𝑚𝑚[𝐽𝐽𝑚𝑚 (𝑥𝑥0 )/𝑥𝑥0 ]2 }. (A6)

*I. S. Gradshteyn and I. M. Ryzhik, “Table of Integrals, Series, and Products,” 7th Edition, Academic Press (2007).

21
 𝑥𝑥 2 (𝑥𝑥) 2 ′ (𝑥𝑥)]𝑑𝑑𝑑𝑑
∫0 0[𝑥𝑥𝐽𝐽𝑚𝑚 − 𝑥𝑥𝐽𝐽𝑚𝑚+1 (𝑥𝑥) − 2𝑚𝑚 𝐽𝐽𝑚𝑚 (𝑥𝑥) 𝐽𝐽𝑚𝑚
= ½𝑥𝑥02 { 𝐽𝐽𝑚𝑚
2 (𝑥𝑥 ) 2 2
0 − 𝐽𝐽𝑚𝑚−1 (𝑥𝑥0 )𝐽𝐽𝑚𝑚+1 (𝑥𝑥0 ) − 𝐽𝐽𝑚𝑚+1 (𝑥𝑥0 ) + 𝐽𝐽𝑚𝑚 (𝑥𝑥0 )𝐽𝐽𝑚𝑚+2 (𝑥𝑥0 ) − 2𝑚𝑚[𝐽𝐽𝑚𝑚 (𝑥𝑥0 )/𝑥𝑥0 ] }

= ½𝑥𝑥02 { 𝐽𝐽𝑚𝑚 (𝑥𝑥0 )[𝐽𝐽𝑚𝑚 (𝑥𝑥0 ) + 𝐽𝐽𝑚𝑚+2 (𝑥𝑥0 )] − 𝐽𝐽𝑚𝑚+1 (𝑥𝑥0 )[𝐽𝐽𝑚𝑚−1 (𝑥𝑥0 ) + 𝐽𝐽𝑚𝑚+1 (𝑥𝑥0 )] − 2𝑚𝑚[𝐽𝐽𝑚𝑚 (𝑥𝑥0 )/𝑥𝑥0 ]2 }
= 𝑥𝑥02 {(𝑚𝑚 + 1)𝐽𝐽𝑚𝑚 (𝑥𝑥0 )𝐽𝐽𝑚𝑚+1 (𝑥𝑥0 )/𝑥𝑥0 − 𝑚𝑚𝐽𝐽𝑚𝑚+1 (𝑥𝑥0 )𝐽𝐽𝑚𝑚 (𝑥𝑥0 )/𝑥𝑥0 − 𝑚𝑚[𝐽𝐽𝑚𝑚 (𝑥𝑥0 )/𝑥𝑥0 ]2 }
= 𝑥𝑥02 { 𝐽𝐽𝑚𝑚 (𝑥𝑥0 )𝐽𝐽𝑚𝑚+1 (𝑥𝑥0 )/𝑥𝑥0 − 𝑚𝑚[𝐽𝐽𝑚𝑚 (𝑥𝑥0 )/𝑥𝑥0 ]2 }
= [𝑥𝑥0 𝐽𝐽𝑚𝑚+1 (𝑥𝑥0 ) − 𝑚𝑚𝐽𝐽𝑚𝑚 (𝑥𝑥0 )] 𝐽𝐽𝑚𝑚 (𝑥𝑥0 ). (A7)
𝑑𝑑𝑑𝑑
� √𝑎𝑎+𝑥𝑥 2 = ln�√𝑎𝑎 + 𝑥𝑥 2 + 𝑥𝑥� (Gradshteyn & Ryzhik 2.261) (A8)

𝑑𝑑𝑑𝑑 𝑥𝑥
� (𝑎𝑎+𝑥𝑥 2 )3⁄2 = (Gradshteyn & Ryzhik 2.271-5) (A9)
𝑎𝑎√𝑎𝑎+𝑥𝑥 2

𝑑𝑑𝑑𝑑 1 𝑥𝑥 𝑥𝑥 3
� (𝑎𝑎+𝑥𝑥 2 )5⁄2 = 𝑎𝑎2 �√𝑎𝑎+𝑥𝑥 2 − 3(𝑎𝑎+𝑥𝑥 2 )3⁄2 � (Gradshteyn & Ryzhik 2.271-6) (A10)

1 1
� 𝑥𝑥√1 − 𝑥𝑥 2 𝑑𝑑𝑑𝑑 = −⅓(1 − 𝑥𝑥 2 )3⁄2 �0 = ⅓ (Gradshteyn & Ryzhik 2.262-2) (A11)
0

𝜋𝜋 −𝜋𝜋𝜋𝜋, (𝑎𝑎2 < 1)

∫0 cos 𝑥𝑥 ln(1 − 2𝑎𝑎 cos 𝑥𝑥 + 𝑎𝑎2 ) 𝑑𝑑𝑑𝑑 = � (Gradshteyn & Ryzhik 4.397-6) (A12)
− 𝜋𝜋⁄𝑎𝑎 , (𝑎𝑎2 > 1)

Appendix B
An alternative method of computing the Livens momentum for the system of Fig.3 begins
by expressing the bound current-density associated with the magnetization distribution as
𝑱𝑱bound (𝒓𝒓, 𝑡𝑡) = 𝜇𝜇0−1 𝜵𝜵 × 𝑴𝑴(𝒓𝒓, 𝑡𝑡), followed by invoking the Biot-Savart law of magnetostatics, as
follows:
∞
𝜇𝜇 𝒓𝒓 − 𝒓𝒓′
𝑩𝑩(𝒓𝒓, 𝑡𝑡) = 4𝜋𝜋0 � 𝑱𝑱bound (𝒓𝒓′ , 𝑡𝑡) × |𝒓𝒓 − 𝒓𝒓′|3 𝑑𝑑𝑥𝑥 ′ 𝑑𝑑𝑦𝑦 ′ 𝑑𝑑𝑧𝑧 ′ . (B1)
−∞

Note that the Bio-Savart law ignores the time-dependence of 𝒎𝒎(𝑡𝑡) by ignoring the effects of
retardation. Integration over the volume of space between the parallel plates of Fig.3 then yields
(Livens) ∞
𝒑𝒑𝑒𝑒𝑒𝑒 (𝑡𝑡) = 𝜀𝜀0 ∭−∞ 𝑬𝑬(𝒓𝒓, 𝑡𝑡) × 𝑩𝑩(𝒓𝒓, 𝑡𝑡)𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
𝑨𝑨 × (𝑩𝑩 × 𝑪𝑪) = (𝑨𝑨 ∙ 𝑪𝑪)𝑩𝑩 − (𝑨𝑨 ∙ 𝑩𝑩)𝑪𝑪
∞
𝜎𝜎 +𝑑𝑑 ⁄2 ∞ ∞ 𝒓𝒓 − 𝒓𝒓′
� × �[𝜵𝜵 × 𝑴𝑴(𝒓𝒓′ , 𝑡𝑡)] × |𝒓𝒓 ′|3 � 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑑𝑑𝑥𝑥 ′ 𝑑𝑑𝑦𝑦 ′ 𝑑𝑑𝑧𝑧 ′
= 4𝜋𝜋0 � ∫𝑥𝑥=−𝑑𝑑⁄2 ∫𝑦𝑦=−∞ ∫𝑧𝑧=−∞ 𝒙𝒙 − 𝒓𝒓
−∞
∞
𝜎𝜎 +𝑑𝑑 ⁄2 ∞ ∞ 𝑥𝑥 − 𝑥𝑥 ′
= 4𝜋𝜋0 � [𝜵𝜵 × 𝑴𝑴(𝒓𝒓′ , 𝑡𝑡)] ∫𝑥𝑥=−𝑑𝑑⁄2 ∫𝑦𝑦=−∞ ∫𝑧𝑧=−∞ |𝒓𝒓 − 𝒓𝒓′ |3 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑑𝑑𝑥𝑥 ′ 𝑑𝑑𝑦𝑦 ′ 𝑑𝑑𝑧𝑧 ′
−∞
∞
𝜎𝜎 +𝑑𝑑 ⁄2 ∞ ∞ 𝒓𝒓 − 𝒓𝒓′
� ∙ [𝜵𝜵 × 𝑴𝑴(𝒓𝒓′ , 𝑡𝑡)] ∫𝑥𝑥=−𝑑𝑑⁄2 ∫𝑦𝑦=−∞ ∫𝑧𝑧=−∞ |𝒓𝒓 ′ |3 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑑𝑑𝑥𝑥 ′ 𝑑𝑑𝑦𝑦 ′ 𝑑𝑑𝑧𝑧 ′
− 4𝜋𝜋0 � 𝒙𝒙 − 𝒓𝒓
−∞
∞
𝜎𝜎0 +𝑑𝑑 ⁄2 ∞ (𝑥𝑥 − 𝑥𝑥 ′ )𝜌𝜌
= � [𝜵𝜵 × 𝑴𝑴(𝒓𝒓′ , 𝑡𝑡)] ∫𝑥𝑥=−𝑑𝑑⁄2 ∫𝜌𝜌=0 [(𝑥𝑥 − 𝑥𝑥 ′ )2 + 𝜌𝜌2 ]3⁄2 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑑𝑑𝑥𝑥 ′ 𝑑𝑑𝑦𝑦 ′ 𝑑𝑑𝑧𝑧 ′
2
−∞

22
 ∞
𝜎𝜎0 +𝑑𝑑 ⁄2 ∞ (𝑥𝑥 − 𝑥𝑥′ )𝜌𝜌
− � [𝜵𝜵 × 𝑴𝑴(𝒓𝒓′ , 𝑡𝑡)]𝑥𝑥 ∫𝑥𝑥=−𝑑𝑑⁄2 ∫𝜌𝜌=0 [(𝑥𝑥−𝑥𝑥 ′ )2 + 𝜌𝜌2 ]3⁄2 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑑𝑑𝑥𝑥 ′ 𝑑𝑑𝑦𝑦 ′ 𝑑𝑑𝑧𝑧 ′
2
−∞
x component of curl Terms containing
∞ � and 𝒛𝒛� vanish
𝒚𝒚
𝜎𝜎0 +𝑑𝑑 ⁄2 (𝑥𝑥 − 𝑥𝑥 ′ )
= � [𝜵𝜵 × 𝑴𝑴(𝒓𝒓′ , 𝑡𝑡)]𝑦𝑦,𝑧𝑧 ∫𝑥𝑥=−𝑑𝑑⁄2 𝑑𝑑𝑑𝑑 𝑑𝑑𝑥𝑥 ′ 𝑑𝑑𝑦𝑦 ′ 𝑑𝑑𝑧𝑧 ′
2 �(𝑥𝑥 − 𝑥𝑥 ′ )2
−∞
y,z components of curl
𝜎𝜎0 ∞ ⁄2
= � [𝜵𝜵 × 𝑴𝑴(𝒓𝒓′ , 𝑡𝑡)]𝑦𝑦,𝑧𝑧 �|𝑥𝑥 − 𝑥𝑥 ′ |+𝑑𝑑 ′ ′
𝑥𝑥=−𝑑𝑑 ⁄2 �𝑑𝑑𝑥𝑥 𝑑𝑑𝑦𝑦 𝑑𝑑𝑧𝑧
′
2 −∞
∞
� + (𝜕𝜕𝑥𝑥 ′ 𝑀𝑀𝑦𝑦 − 𝜕𝜕𝑦𝑦 ′ 𝑀𝑀𝑥𝑥 )𝒛𝒛��𝑥𝑥 ′ 𝑑𝑑𝑥𝑥 ′ 𝑑𝑑𝑦𝑦 ′ 𝑑𝑑𝑧𝑧 ′
= −𝜎𝜎0 � �(𝜕𝜕𝑧𝑧 ′ 𝑀𝑀𝑥𝑥 − 𝜕𝜕𝑥𝑥 ′ 𝑀𝑀𝑧𝑧 )𝒚𝒚 Integration by parts
−∞
∞ ∞
= 𝜎𝜎0 � 𝑥𝑥 ′ 𝜕𝜕𝑥𝑥 ′ �𝑀𝑀𝑧𝑧 𝒚𝒚
� − 𝑀𝑀𝑦𝑦 𝒛𝒛��𝑑𝑑𝑥𝑥 ′ 𝑑𝑑𝑦𝑦 ′ 𝑑𝑑𝑧𝑧 ′ = −𝜎𝜎0 � �𝑀𝑀𝑧𝑧 𝒚𝒚
�− 𝑀𝑀𝑦𝑦 𝒛𝒛��𝑑𝑑𝑥𝑥 ′ 𝑑𝑑𝑦𝑦 ′ 𝑑𝑑𝑧𝑧 ′
−∞ −∞
∞
= � × ∭−∞ 𝑴𝑴(𝒓𝒓′ , 𝑡𝑡)𝑑𝑑𝑥𝑥 ′ 𝑑𝑑𝑦𝑦 ′ 𝑑𝑑𝑧𝑧 ′ = 𝜀𝜀0 𝑬𝑬0 × 𝒎𝒎(𝑡𝑡).
𝜎𝜎0 𝒙𝒙 (B2)

The above result is the same as that used in Eq.(53). While the Abraham momentum of the
EM field in the system of Fig.3 is equal to zero, the Livens momentum has a contribution arising
from the interaction between the external 𝐸𝐸-field and the magnetic dipole moment of the particle.

References
1. M. Mansuripur, “The Force Law of Classical Electrodynamics: Lorentz versus Einstein and Laub,” Proc. of
SPIE Vol. 8810, 8810OK-1:18 (2013); available online at <arXiv:1312.3262>.
2. M. Mansuripur, “Electromagnetic force and torque in Lorentz and Einstein-Laub formulations,” Proc. SPIE Vol.
9164, 91640B~1:16 (2014); available online at <arXiv:1409.5860>.
3. M. Mansuripur, “Electric and Magnetic Dipoles in the Lorentz and Einstein-Laub Formulations of Classical
Electrodynamics,” Proc. SPIE Vol. 9370, 93700U~1:15 (2015); available online at <arXiv:1503.02111>.
4. J. D. Jackson, Classical Electrodynamics, 3rd edition, Wiley, New York, 1999.
5. D.J. Griffiths, Introduction to Electrodynamics, 2nd edition, Prentice Hall, New Jersey, 1989.
6. R. P. Feynman, R. B. Leighton, and M. Sands, The Feynman Lectures on Physics, Addison-Wesley, Reading,
Massachusetts (1964).
7. R. M. Fano, L. J. Chu, and R. B. Adler, Electromagnetic Fields, Energy and Forces, Wiley, New York (1960).
8. M. Mansuripur, Field, Force, Energy and Momentum in Classical Electrodynamics, Bentham e-books (2011).
9. W. Shockley and R.P. James, “Try simplest cases discovery of hidden momentum forces on magnetic currents,”
Phys. Rev. Lett. 18, 876-879 (1967).
10. W. Shockley, “Hidden linear momentum related to the α ·E term for a Dirac-electron wave packet in an electric
field,” Phys. Rev. Lett. 20, 343-346 (1968).
11. S. Coleman and J. H. Van Vleck, “Origin of ‘hidden momentum forces’ on magnets,” Phys. Rev. 171, 1370-75
(1968).
12. H. A. Haus and P. Penfield, “Force on a current loop,” Phys. Lett. 26A, 412-413 (1968).
13. O. Costa de Beauregard, “‘Hidden momentum’ in magnets and interaction energy,” Phys. Lett. 28A, 365 (1968).
14. W. H. Furry, “Examples of momentum distribution in the electromagnetic field and in matter,” Am. J. Phys. 37,
621-36 (1969).
15. L. Vaidman, “Torque and force on a magnetic dipole,” Am. J. Phys. 58, 978-983 (1990).
16. D. Bedford and P. Krumm, “On the origin of magnetic dynamics,” Am. J. Phys. 54, 1036 (1986).
17. V. Namias, “Electrodynamics of moving dipoles: The case of the missing torque,” Am. J. Phys. 57, 171-177
(1989).
18. D. Bedford and P. Krumm, “Comment on electrodynamics of moving dipoles: The case of the missing torque,”
Am. J. Phys. 57, 178 (1989).
19. V. Namias, “A discussion of the dielectric model of Bedford and Krumm,” Am. J. Phys. 57, 178-179 (1989).
20. G. Spavieri, “Proposal for experiments to detect the missing torque in special relativity,” Foundations of Physics
Letters 3, 291-302 (1990).

23
21. V. Hnizdo, “Comment on ‘Torque and force on a magnetic dipole’,” Am. J. Phys. 60, 279-280 (1992).
22. E. Comay, “Exposing hidden momentum,” Am. J. Phys. 64, 1028-1034 (1996).
23. V. Hnizdo, “Hidden momentum of a relativistic fluid carrying current in an external electric field,” Am. J. Phys.
65, 92-94 (1997).
24. T. H. Boyer, “Concerning ‘hidden momentum’,” Am. J. Phys. 76, 190-191 (2008).
25. D. Babson, S. P. Reynolds, R. Bjorkquist, and D. J. Griffiths, “Hidden momentum, field momentum, and
electromagnetic impulse,” Am. J. Phys. 77, 826-833 (2009).
26. D. J. Griffiths, “Resource Letter EM-1: Electromagnetic Momentum,” Am. J. Phys. 80, 7-18 (2012).
27. M. Mansuripur, “Trouble with the Lorentz Law of Force: Incompatibility with Special Relativity and
Momentum Conservation,” Phys. Rev. Lett. 108, 193901 (2012); available online at <arXiv:1205.0096>.
28. M. Mansuripur, “Trouble with the Lorentz Law of Force: Response to Critics,” Proc. SPIE Vol. 8455,
845512~1-13 (2012); available online at <arXiv:1211.3485>.
29. D. J. Griffiths and V. Hnizdo, “Mansuripur’s paradox,” Am. J. Phys. 81, 570-574 (2013).
30. D. J. Griffiths and V. Hnizdo, “What’s the Use of Bound Charge?”, available online at <arXiv:1506.02590>.
31. M. Mansuripur, “On the Foundational Equations of the Classical Theory of Electrodynamics,” Resonance 18(2),
130-155 (2013); available online at <arXiv:1404.2526>.
32. M. Mansuripur, “The Lorentz force law and its connections to hidden momentum, the Einstein-Laub force, and
the Aharonov-Casher effect,” IEEE Trans. Magnet. 50, 1300110 (2014); available online at <arXiv:1404.3261>.
33. A. Einstein and J. Laub, "Über die im elektromagnetischen Felde auf ruhende Körper ausgeübten pondero-
motorischen Kräfte," Annalen der Physik 331, 541–550 (1908). The English translation of this paper appears in
volume 2 of The Collected Papers of Albert Einstein (Princeton University Press, Princeton, NJ, 1989).
34. B. A. Kemp, “Resolution of the Abraham-Minkowski debate: Implications for the electromagnetic wave theory
of light in matter,” J. Appl. Phys. 109, 111101 (2011).

24