Вы находитесь на странице: 1из 10

Faraday induction and the current carriers in a circuit

Timothy H. Boyer

Citation: American Journal of Physics 83, 263 (2015); doi: 10.1119/1.4901191

View online: http://dx.doi.org/10.1119/1.4901191
View Table of Contents: http://scitation.aip.org/content/aapt/journal/ajp/83/3?ver=pdfcov
Published by the American Association of Physics Teachers

Articles you may be interested in

The electromagnetic momentum of static charge-current distributions
Am. J. Phys. 82, 869 (2014); 10.1119/1.4879539

Mutual inductance between piecewise-linear loops

Am. J. Phys. 81, 829 (2013); 10.1119/1.4818278

Teaching Faraday’s law of electromagnetic induction in an introductory physics course

Am. J. Phys. 74, 337 (2006); 10.1119/1.2180283

Challenges to Faraday’s flux rule

Am. J. Phys. 72, 1478 (2004); 10.1119/1.1789163

The Flexible Faraday Cage

Phys. Teach. 42, 181 (2004); 10.1119/1.1664386

This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: On: Tue, 24 Feb 2015 15:35:23
Faraday induction and the current carriers in a circuit
Timothy H. Boyer
Department of Physics, City College of the City University of New York, New York, New York 10031
(Received 16 August 2014; accepted 27 October 2014)
This article treats Faraday induction from an untraditional, particle-based point of view. The
electromagnetic fields of Faraday induction can be calculated explicitly from approximate
point-charge fields derived from the Lienard–Wiechert expressions, or from the Darwin
Lagrangian. Thus the electric fields of electrostatics, the magnetic fields of magnetostatics, and the
electric fields of Faraday induction can all be regarded as arising from charged particles. Some
aspects of electromagnetic induction are explored for a hypothetical circuit consisting of point
charges that move frictionlessly in a circular orbit. For a small number of particles in the circuit
(or for non-interacting particles), the induced electromagnetic fields depend upon the mass and
charge of the current carriers while energy is transferred to the kinetic energy of the particles.
However, for an interacting multiparticle circuit, the mutual electromagnetic interactions between
the particles dominate the behavior so that the induced electric field cancels the inducing force per
unit charge, the mass and charge of the individual current carriers become irrelevant, and energy
goes into magnetic energy. VC 2015 American Association of Physics Teachers.


When students are asked what causes the electric field in a A. Point-charge fields for general motion
parallel-plate capacitor, the response involves charges on the
capacitor plates. Also, students say that the magnetic field in Although students are familiar with the Coulomb electric
a solenoid is due to the currents in the solenoid winding. But field of a point charge, many do not study electromagnetism
when asked for the cause of the Faraday induction field in a to the point that they see the full retarded point-charge fields
solenoid with changing currents, the usual student response from the Lienard–Wiechert potentials6
is that the induction field is due to a changing magnetic "  #
field—not that it is due to the acceleration fields of the 1  v2e =c2 ðn  ve =cÞ
Eðr; tÞ ¼ e
charges in the solenoid winding. The student view reflects ð1  n  ve =cÞ3 jr  re j2 t
what is emphasized in the standard electromagnetism text- " #
books.1,2 Indeed, although Darwin3 computed induction e n  fðn  ve =cÞ  ae g
þ 2 ; (1)
fields from accelerating charges in the 1930s, today it is rare c ð1  n  ve =cÞ3 jr  re j
to have a physicist report that the Faraday induction field tret
arises from the acceleration of charges.4,5 and
In this article, we wish to broaden the perspective on
Faraday induction by reviewing some aspects of the particle Bðr; tÞ ¼ ntret  Eðr; tÞ; (2)
point of view. We treat the induction fields as arising from the
electromagnetic fields of point charges as derived from the where the unit vector n ¼ ðr  re Þ=jr  re j, and where the
Lienard–Wiechert expressions or from the Darwin position re ðtÞ, velocity ve ðtÞ, and acceleration ae ðtÞ of the
Lagrangian. First, we mention the Lienard–Wiechert6 form charge e must be evaluated at the retarded time tret such that
taken by the electric and magnetic fields of a point charge in cjt  tret j ¼ jr  re ðtret Þj. These field expressions, together
general motion. Then we turn to the low-velocity-small-dis- with Newton’s second law for the Lorentz force, give the
tance approximation derived in the 1940 textbook by Page causal interactions between point charges.9 The only thing
and Adams.7 This approximate form for the electromagnetic missing from this formulation of classical electrodynamics is
fields is the same as that obtained from the Darwin the possible existence of a homogeneous solution of
Lagrangian8 of 1920. Here in the present article, the approxi- Maxwell’s equations (such as, for example, a plane wave)
mated fields are used to treat Faraday induction in detail for a that might interact with the charged particles. However, elec-
hypothetical circuit consisting of point charged particles mov- tromagnetic induction fields are distinct from homogeneous
ing frictionlessly on a circular ring. This circuit provides a radiation fields, and therefore, we expect that the fields of
rough approximation to that of a thin wire that is bent into a Faraday induction can be treated as having their origin from
circular loop. For small numbers of particles (or for noninter- charged particle motions. The exploration of this point-
acting particles), we see that the magnitudes of the masses charge point of view in connection with Faraday induction is
and charges of the charge carriers are important, that the the subject of the present article.
induced electric fields can be small, and that energy goes into
the mechanical kinetic energy of the charge carriers. B. Low-velocity-small-distance approximation without
However, for large numbers of interacting charges, the mutual retardation
interactions make the magnitudes of the masses and charges
unimportant, the induced electric fields balance the inducing Since the electric field of Faraday induction is distinct
fields, and the energy goes into magnetic energy of the circuit. from the Coulomb field of a stationary charge, we expect

263 Am. J. Phys. 83 (3), March 2015 http://aapt.org/ajp C 2015 American Association of Physics Teachers
V 263

This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: On: Tue, 24 Feb 2015 15:35:23
" #    
this induction field to involve the additional velocity- and d mi v i d v2i
acceleration-dependent terms of Eq. (1). The use of the full    mi 1 þ 2 v i
dt 1  v2 =c2 1=2 dt 2c
expressions involving retardation in Eqs. (1) and (2) can be a i
formidable task. For the radiation fields, which fall off as vi
1=r at large distances, the presence of retardation cannot be ¼ ei E þ ei  B
 c X 
avoided as the signal travels from the source charge to the
¼ ei Eext ðri ;tÞ þ Ej ðri ;tÞ
distant field point. However, for field points that are near
point charges that are moving at low velocities, it is possible
to derive from Eqs. (1) and (2) approximate expressions vi  X 
þ ei  Bext ðri ;tÞ þ Bj ðri ;tÞ ;
for the electric and magnetic fields that involve no retarda- c j6¼i
tion. The task of approximation is not trivial and is carried
out in the textbook Electrodynamics by Page and Adams7
giving with the Lorentz force on the ith particle arising from the
"  # external electromagnetic fields and from the electromag-
ðr  r j Þ v2j 3 vj  ðr  rj Þ 2 netic fields of the other particles. The electromagnetic fields
Ej ðr;tÞ ¼ej 1þ 2 
jr  rj j3 2c 2 cjr  rj j due to the jth particle are given through order v2 =c2 by
" # exactly the approximate expressions appearing in Eqs. (3)
ej aj aj  ðr  rj Þðr  rj Þ  
 2 þ þ O 1=c3 ; and (4).
2c jr  rj j 3
jr  rj j
and Electromagnetic induction was discovered by Michael
  Faraday, not as a motion-dependent modification of
vj ðr  rj Þ 3
Bj ðr; tÞ ¼ ej  þ O 1=c ; (4) Coulomb’s law, but rather in terms of emfs producing
c jr  rj j3 currents in circuits. This circuit-based orientation remains
the way that electromagnetic induction is discussed in text-
where in Eq. (3) the quantity aj refers to the acceleration of books today. The emf in a circuit is the closed line integral
the jth particle. These approximate expressions are power around the circuit of the force per
Þ unit charge f acting on the
series in 1=c, where c is the speed of light in vacuum. The charges of the circuit: emf ¼ f  dr. Faraday’s emfs were
approximate fields in Eqs. (3) and (4) can be quite useful; associated with changing magnetic fluxes, and Faraday’s law
they were used by Page and Adams10 to discuss “action and of electromagnetic induction in a circuit is given by
reaction between moving charges,” and by the current author
when interested in Lorentz-transformation properties of 1 dU
emf F ¼  ; (7)
energy and momentum11 and questions of mass-energy c dt
equivalence.12 Here, the approximate expressions for the
fields are exactly what is needed to understand Faraday where U is the magnetic flux through the circuit.
induction from a particle point of view. As correctly emphasized in some textbooks,13 electromag-
The approximate fields in Eqs. (3) and (4) also correspond netic induction in a circuit can arise in two distinct aspects.
to those that arise from the Darwin Lagrangian8 The “motional emf” in a circuit that is moving through an
unchanging magnetic field can be regarded as arising from
 2 2 ! the magnetic Lorentz force acting on the mobile charges of
v 2 v 1Xi¼N X
ei ej
L¼ mi c2 1 þ i2 þ i 4  the moving circuit. On the other hand, when the circuit is sta-
2c 8c 2 i¼1 j6¼i jri  rj j tionary in space but the current in the circuit is changing,
" # new electric fields arise. These new electric fields can cause
1Xi¼N X
ei ej vi  vj vi  ðri  rj Þvj  ðri  rj Þ an emf in an adjacent circuit (mutual inductance) or in the
þ þ
2 i¼1 j6¼i 2c2 jri  rj j jri  rj j3 original circuit itself (self-inductance); the new electric fields
are precisely those appearing due to the motions of the
i¼N X
vi charges of the circuit as given in Eq. (1), or, through order
 ei Uext ðri ; tÞ þ ei  Aext ðri ; tÞ; (5) v2 =c2 , as given in Eq. (3). It is these electric fields that are
i¼1 i¼1
the subject of our discussion of Faraday induction.
It should be noted that for steady-state currents in a multi-
where the last line includes the scalar potential Uext and particle circuit with large numbers of charges where the
vector potential Aext associated with the external electromag- charge density and current density are time-independent, all
netic fields. The Darwin Lagrangian omits radiation but the complicating motion-dependent terms in Eqs. (1)–(2) or
expresses accurately the interaction of charged particles (3)–(4) beyond the first leading term in 1=c actually cancel,
through order 1=c2 . The Darwin Lagrangian continues to so that the electromagnetic fields can be calculated simply
appear in advanced textbooks,8 but the approximate expres- using Coulomb’s law and the Biot-Savart Law.14 However,
sions (3) and (4) seem to have disappeared from the con- for time-varying charge densities and/or current densities,
sciousness of most contemporary physicists. The Lagrangian the motion-dependent terms in Eqs. (1)–(2) or (3)–(4) do not
equations of motion from the Darwin Lagrangian can be cancel and indeed provide the Faraday induction fields.
rewritten in the form of Newton’s second law dp=dt ¼ If an external emf, emf ext , is present in a continuous cir-
dðmcvÞ=dt ¼ F with c ¼ ð1  v2 =c2 Þ1=2 . In this Newtonian cuit with a self-inductance L and resistance R, the current i
form, we have in the circuit is given by the differential equation

264 Am. J. Phys., Vol. 83, No. 3, March 2015 Timothy H. Boyer 264

This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: On: Tue, 24 Feb 2015 15:35:23
di analysis to follow. The system can be thought of as consist-
emf ext ¼ L þ iR; (8)
dt ing of charged beads sliding on a frictionless ring. There is
no frictional force and hence no resistance R in the model.
where the term L di=dt corresponds to the negative of the The model is intended as a rough approximation to a circular
Faraday-induced emf associated with the changing current in loop of wire of small cross section.15
the circuit. Here, in traditional electromagnetic theory, the We now imagine that a constant external force per unit
self-inductance L of a rigid circuit is a time-independent ^
charge f ext is applied in a circular pattern in the tangential /
quantity that depends only upon the geometry of the circuit. ^
direction, f ext ¼ /fext , to all the charges of the ring. One
The energy balance for the circuit is found by multiplying need not specify the source of f ext , but one example would
Eq. (8) by the current i be an axially symmetric magnetic field applied perpendicular
  to the plane of the circular orbit in the ^z direction, increas-
d 1 2 ing in magnitude at a constant rate. The external emf around
emf ext  i ¼ Li þ i2 R; (9)
dt 2 the circular orbit is
corresponding to a power emf ext  i delivered by the exter- emf ext ¼ f ext  dr ¼ 2pRfext : (10)
nal emf going into the time-rate-of-change of magnetic
energy ð1=2ÞLi2 stored in the inductor and the power i2 R
lost in the resistor. The external force per unit charge f ext places a tangential
Although the energy analysis for Eq. (9) seems natural, ^ fext on the ith particle located at ri . The
force Fi ¼ ei / i
the differential equation (8) presents some unusual aspects if Faraday inductance of the charged-particle system is deter-
we consider the circuit from the particle point of view. If at mined by the response of all the particles ei in the circular
time t ¼ 0, the constant external emf (emf ext ) is applied to orbit.
the circuit and the current is zero (ið0Þ ¼ 0) then the Faraday
induced emf (emf F ¼ L di=dt) must exactly cancel emf ext B. One-particle model for a circuit
so that emf ext  Ldi=dt ¼ 0 at time t ¼ 0. Phrased in terms
of forces per unit charge applied to the circuit, the Faraday 1. Motion of the charged particle
induced electric field EF must exactly cancel the external
We start with the case when there is only one charged par-
force per unit charge f ext associated with the external emf.
ticle of mass m and charge e in the circular orbit. In this
Indeed, if the resistance R of the circuit becomes vanish-
case, the tangential acceleration a/ of the particle arises
ingly small (R ! 0) then this canceling balance of the
from the (tangential) force of only the external force per unit
Faraday electric field EF against the force per unit charge
charge f ext , since the centripetal forces of constraint are all
f ext associated with the external emf holds at all times, and
radial forces. From Eq. (6), written for a single particle and
yet the current increases at a constant rate following di=dt ¼
with dðmcvÞ=dt ¼ mc3 a/ where c ¼ ð1  v2 =c2 Þ1=2 , we
emf ext =L: But if the net force per unit charge goes to zero,
why do the charges accelerate so as to produce a changing
current di=dt? After all, in classical mechanics it is the re- efext
sultant force FR on a particle that determines the particle’s a/ ¼ ; (11)
acceleration a; that is, FR ¼ ma. Thus, we expect that if the mc3
resultant force on a particle is zero, then the particle does not
where fext is the magnitude of the tangential force per unit
accelerate. However, electromagnetism involves some
charge due to the external emf (emf ext ) at the position of the
aspects that are different from what is familiar in nonrelativ-
charge e.
istic mechanics, and electromagnetic circuit theory involves
some approximations that go unmentioned in the textbooks.
In this article, we wish to explore these differences and 2. Magnetic field of the charged particle
unmentioned approximations by using a particle model in
The magnetic field Be at the center of the circular orbit
connection with Faraday induction. We will note the approx-
due to the accelerating charge e is given by Eq. (4)
imations involved in Eq. (8) that lead to the troubling appa-
rent contradiction with Newton’s second law. ^ v ;
Be ð0; tÞ ¼ ke (12)
IV. FARADAY INDUCTION IN A SIMPLE where the velocity v is increasing since the external force per
HYPOTHETICAL CIRCUIT unit charge f ext gives a positive charge e a positive accelera-
tion in the /^ direction. This magnetic field Be produced by
A. Model for a detailed discussion
the orbiting charge e is increasing in the ^z direction, which is
Here, we would like to explore Faraday induction in some in the opposite direction from the increasing external mag-
detail for the simplest possible circuit, in hopes of obtaining netic field that could have created f ext and emf ext in Eq. (10).
some physical insight. Accordingly, we will discuss a hypo-
thetical circuit consisting of N equally-spaced particles of 3. Induced electric field from Faraday’s law
mass m and charge e, which are constrained by centripetal
forces to move in a circular orbit of radius R in the xy-plane, Associated with this changing magnetic field Be , created
centered on the origin. A balancing negative charge to make by the orbiting charge e, there should be an induced electric
the circuit electrically neutral can be thought of as a uniform field Ee ðr; tÞ according to Faraday’s law. Thus averaging
line charge in the orbit, or as a single compensating charge over the circular motion of the charge, we expect an average
at the center of the orbit; the choice does not influence the induced tangential electric field hEe/ ðrÞi at a distance r from

265 Am. J. Phys., Vol. 83, No. 3, March 2015 Timothy H. Boyer 265

This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: On: Tue, 24 Feb 2015 15:35:23
the center of the circular orbit (where r  R so that the due to the charge e obtained by use of the approximate elec-
magnetic field Be has approximately the value Bð0; tÞ at the tric field expression given in Eq. (3). Thus we assume that
center) given from Eq. (12) by the charge e is located momentarily at re ¼ x^R cos /e
þ^ y R sin /e , and we average the electric field Ee ðr; tÞ due to e
1 dUe 1d
2prhEe/ ðr Þi ¼ emf e ¼  ¼ Be ð0; tÞpr2 over the phase /e . Since the entire situation is axially sym-
 c dt  c dt  metric when averaged over /e , we may take the field point
1d v 2 1 a/ along the x-axis at r ¼ x^r, and later generalize to cylindrical
¼ e 2 pr ¼  e 2 pr2 ;
c dt cR c cR coordinates. The velocity fields given in the first line of
(13) Eq. (3) point from the charge e to the field point. Also, the
velocity fields are even if the sign of the velocity ve is
since dv=dt ¼ a/: Using Eq. (11), the average tangential changed to ve . Thus the velocity fields when averaged over
electric field follows from Eq. (13) as the circular orbit can point only in the radial direction. The
acceleration fields arising from the centripetal acceleration
^ e2 rfext of the charge will also point in the radial direction. Since we
hEe/ ðr; tÞi ¼ / : (14)
2mc2 c3 R2 are interested in the average tangential component of the
field Ee , we need to average over only the tangential acceler-
This derivation of the Faraday induction field corresponds to ation terms in the second line of Eq. (3). If the field point is
the familiar textbook approach. close to the center of the circular orbit so that r  R, then
we may expand in powers of r=R; we retain only the first-
4. Induced electric field from the approximate point-charge order terms, giving j^ x r  re j1  R1 ð1 þ x^r  re =R2 Þ and
fields 3
x r  re j  R3 ð1 þ 3^
j^ x r  re =R2 Þ. Then the average tan-
We will now show that this induced average tangential gential component of the electric field due to the charge e
electric field hEe/ ðr; tÞi is exactly the average electric field can be written as

* " #+
e ae/ ae/  ðx^r  re Þðx^r  re Þ
hEe/ ðx^r; tÞi ¼  2 þ
2c j^x r  re j x r  re j 3

e ae/ x^r  re ae/  ðx^r  re Þðx^r  re Þ x r  re
¼  2 1þ þ 1þ : (15)
2c R R2 R3 R2

Now we average over the phase /e with re ¼ x^R cos /e þ the other hand, if we try to increase the induced electromag-
y^R sin /e and ae/ ¼ ae/ ð^
x sin /e þ y^ cos /e Þ. We note that netic field Ee/ by making the mass m small, we encounter a
hae/ i ¼ 0; ae/  re ¼ 0, and hae/ ð^x  re Þi ¼ y^ae/ R=2 fundamental limit of electromagnetic theory. The allowed
¼ hðae/  x^Þre i. After averaging and retaining terms mass m is limited below by considerations involving the
through order r/R, Eq. (15) becomes classical radius of the electron rcl ¼ e2 =ðmc2 Þ. Classical
electromagnetic theory is valid only for distances large com-
ea/ r pared to the classical radius of the electron. Thus in our
hEe/ ð^
x r; tÞi ¼ ^
y ; (16)
2c2 R2 example where the radius R of the orbit is a crucial parame-
ter, we must have R  rcl . This means we require the mass
which is in agreement with our earlier results in Eqs. (13) m  e2 =ðRc2 Þ and so e2 =ðmc2 RÞ  1. Combining this limit
and (14). Thus indeed the electric field of Faraday induction with r=R < 1 and 1 < c leads to a limit on the magnitude of
in this case arises from the acceleration of the charged cur- the induced electric field in Eq. (14)
rent carrier of the circuit.
hEe/ ðr; tÞi  fext for r < R: (17)
5. Limit on the induced electric field
The induced electric field of a one-particle circuit is small
We are now in a position to comment on the average compared to the external force per unit charge associated
response of our one-particle circuit to the applied external with the external emf.
emf. If the source of emf ext is a changing magnetic field,
then this situation corresponds to the traditional example for
diamagnetism within classical electromagnetism.16 For this 6. Energy balance
one-particle example, the response depends crucially upon
the mass m and charge e of the particle. When the mass m is We also note that the power delivered by the external
large, the acceleration of the charge is small; therefore the force per unit charge goes into kinetic energy of the orbiting
induced tangential electric field Ee/ in Eq. (14) is small. particle. Thus if we take the Newton’s-second-law equation
This large-mass situation is what is usually assumed in giving Eq. (11) and multiply by the speed v of the particle,
examples of charged rings responding to external emfs.17 On we have

266 Am. J. Phys., Vol. 83, No. 3, March 2015 Timothy H. Boyer 266

This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: On: Tue, 24 Feb 2015 15:35:23
d d  consistent with the remaining terms arising from the approxi-
ðmcvÞv ¼ mcc2 ¼ mc3 a/ v ¼ efext v; (18)
dt dt mate field expression (3).
Since the particles are equally spaced around the circular
so that the power efext v delivered to the charge e by the exter- orbit and all have the same charge e and mass m, the situa-
nal force goes into kinetic energy of the particle. tion is axially symmetric. The equation of motion for every
The situation of a one-particle circuit can be summar- charge takes the same form, and the angular acceleration of
ized as follows. For the one-particle circuit, the induced each charge is the same: d2 /i =dt2 ¼ d 2 /=dt2 . For simplicity
electric field is small compared to the external force per of calculation, we will take the Nth particle along the x-axis
unit charge and depends explicitly upon the particle’s so that /N ¼ 0; rN ¼ x^R, and / ^ ¼ y^. The other particles
mass and charge, while the energy transferred by the are located at rj ¼ x^R cosð2pj=NÞ þ y^R sinð2pj=NÞ, corre-
external field goes into kinetic energy of the one charged sponding to an angle /j ¼ 2pj=N for j ¼ 1; 2; …; N  1. The
particle. Clearly, this is not the situation that we usually tangential acceleration of the jth particle is given by
associate with electromagnetic induction for circuit aj/ ¼ ðd 2 /=dt2 Þ½^
x R sinð2pj=NÞ þ y^R cosð2pj=NÞ. By
problems. symmetry, it is clear that the electrostatic fields, the velocity
fields, and the centripetal acceleration fields of the other par-
ticles cannot contribute to the tangential electric field at par-
C. Multiparticle model for a circuit ticle N. The equation of motion for the tangential
acceleration for each charge in the circular orbit is the same
1. Motion of the charged particles as that for the Nth particle, which from Eq. (19) is
In order to make contact with the usual discussion of 8
N 1 2 
Faraday inductance in a circuit, we must go to the situation d / <
2 X e y^  aj/
mc3 R 2 ¼ efext  2 j^
of many particles, each of charge e and mass m. However, dt : j¼1
2c x R  rj j
if we take the one-particle circuit above and simply super- #)
impose the fields corresponding to N equally spaced aj/  ð x^R  rj Þy^  ð x^R  rj Þ
charges while maintaining the acceleration appropriate for þ : (20)
x R  rj j3
the single-particle case, we arrive at a completely false
result. If we take Eq. (14) for the Faraday-induced average
electric field hEe/ ðr; tÞi due to a single particle of charge e Now we evaluate the distance between the jth particle and
and mass m accelerating in the external force per unit the Nth particle in the circular orbit as
charge fext and then simply multiply by the number of
charges, we have a result that is linear in N and increases x R  rj j ¼ ½2R2  2R2 cosð2pj=NÞ1=2
without bound. Thus, merely extrapolating from the one-
¼ ½4R2 sin2 ðpj=NÞ1=2 ¼ j2R sinðpj=NÞj; (21)
particle circuit suggests that the Faraday induced electric
field arising from many charges might far exceed the
inducing force per unit charge fext .
To obtain a correct understanding of the physics, we
y^  aj/ ¼ ðd2 /=dt2 ÞR cosð2pj=NÞ (22)
must include the mutual interactions between the accelerat-
ing charges of the circuit. With these mutual interactions,
the force on any charge in the circular orbit is not just the
external force efext due to the original external force per aj  ð^
x R  rj Þ^
y  ð^
x R  rj Þ
unit charge, and the acceleration of any charge is not given
by a/ ¼ efext =ðmc3 Þ. Now the force on any charge is a sum ¼ ðd2 /=dt2 ÞR½R sinð2pj=NÞ½R sinð2pj=NÞ: (23)
of the force due to the original external force per unit
charge plus the forces due to the fields of all the other Then Eq. (20) becomes
charged particles in the circular orbit as given in Eq. (6). d2 /
The magnetic force ei vi  B=c is simply a deflection and mc3 R
does not contribute to the tangential acceleration of the N1 2
charge ei. Thus the equation of motion for the ith particle d2 / X e R cosð 2pj=N Þ
¼ efext  2
becomes dt j¼1 2c2 j2R sinð pj=N Þj
d ^ R½ R sinð 2pj=N Þ½ R sinð 2pj=N Þ
ðmi ci vi Þ  / þ
dt j2R sinð pj=N Þj3
d2 / " #
¼ mi c3i ai/ ¼ mi c3i R 2 i d2 / X
N 1 2
e 2  3 sin2 ð pj=N Þ
( dt ¼ efext  2 (24)
X ðr i  r j Þ X e j dt j¼1 2c2 2 sinð pj=N Þ
¼/^  ei f ext ðri Þ þ ej 
j6¼i jri  rj j3 j6¼i
or, solving for d2 /=dt2 ,
" #)
aj a j  ðr i  r j Þðr i  r j Þ 8 " #91
 þ ; (19) < =
jri  rj j jri  rj j3 d2 / X
N1 2
e 2  3 sin 2
ð pj=N Þ
¼ ef ext mc R þ :
dt2 : j¼1
2c2 2 sinð pj=N Þ ;
where it is understood that the factor ci ¼ ð1  v2i =c2 Þ1=2
should be expanded through second order in vi =c so as to be (25)

267 Am. J. Phys., Vol. 83, No. 3, March 2015 Timothy H. Boyer 267

This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: On: Tue, 24 Feb 2015 15:35:23
If there is only one particle on the frictionless ring so that rate of change of a particle’s kinetic energy equals the differ-
N ¼ 1, the sum disappears and the tangential acceleration ence between the power delivered to the particle by the
corresponds to the result obtained earlier in Eq. (11) above, external force and the power absorbed from the particle as
with Rd 2 /=dt2 ¼ a/ . We note that the mass term in Eq. (25) magnetic field energy. Since the acceleration of the particle
remains unchanged by the number of particles while the sum is very small, the rate of change of the particle kinetic energy
increases with each additional particle. Thus, if there are is very small and most of the energy delivered by the exter-
many particles, then the electric field at particle i due to the nal force goes into magnetic field energy. If one neglects the
other particles j can lead to so large a sum in Eq. (25) that very small amount of energy going into particle kinetic
the mass contribution mc3 R becomes insignificant. In this energy, then one simply states that the energy delivered by
case, the common angular acceleration of each particle the external force goes into magnetic field energy. This last
becomes, from Eq. (25), statement is the approximation that appears in the textbooks
8 " of electromagnetism.
#91 In nonrelativistic classical mechanics (which underlies the
d2 / 2c2 <X N1
2  3 sin2 ð pj=N Þ = intuition of most physicists), we can store potential energy of
 fext : (26)
dt2 e : j¼1 2 sinð pj=N Þ ; relative position but there is no such thing as potential
energy of velocity. In classical mechanics, the energy associ-
We see that in this multiparticle situation the angular accel- ated with velocity is always mechanical kinetic energy, mass
eration no longer depends upon the mass m of the charge car- times velocity squared. In complete contrast to this situation,
riers. This is the situation envisioned in the usual textbook electromagnetic theory contains magnetic energy associated
treatment of Faraday induction. with the velocities of charges and yet not associated with
particle mass. For consistency, electromagnetic theory
requires that accelerating charges cause fields that produce
2. Induced electric field forces on other charges. The mutual interaction of the
charges through the Faraday induction fields assures that the
Furthermore, in this multiparticle situation where the magnetic energy stored indeed requires work by some exter-
particle mass becomes insignificant,P the left-hand side of nal forces. For a multiparticle system, the resultant force on
Eq. (20) is negligible, so that the sum j Eej ðri Þ of the accel- each particle may become tiny compared to the external
eration fields of all the other charges ej cancels the external force on each charge, because of the acceleration fields due
force per unit charge f ext ðri Þ of the external emf at the posi- to the other charges. The charges have a tiny acceleration
tion ri of each charge in the circular orbit and gain a tiny amount of kinetic energy while the work
X done by the external force goes into the large amount of
f ext ðri Þ  Ee ðri Þ ¼ Eej ðri Þ: (27) energy stored in the magnetic field.

This situation is analogous to that in electrostatics where the

3. Self-inductance of the circuit
fields of the charges in a conductor move to new positions so
as to cancel the external force per unit charge at the position The value for the self-inductance of the multiparticle cir-
of each charge; here, the charges accelerate so that the accel- cuit can be obtained from Eq. (8) when the circuit resistance
eration fields cancel the external force per unit charge at the vanishes. If the resistance vanishes, the external emf,
position P of each charge. Now the induced electric field emf ext ¼ 2pRfext , around the ring equals the self-inductance
Ee ðrÞ ¼ i Eei ðrÞ at a general field point due to the orbit L multiplied by the time rate of change of the current,
particles is independent of the charge e of the charge carriers, i ¼ Neðd/=dtÞ=ð2pÞ,
since the angular acceleration in Eq. (26) depends inversely  
on the charge e, and this inverse dependence upon e cancels di Ne d 2 /
with the e appearing in Eq. (3) so as to give an induced elec- emf ext ¼ 2pREext ðRÞ ¼ L ¼ L : (28)
dt 2p dt2
tric field that is independent of the charge on the charge car-
riers. Again, this situation corresponds to that treated in the Traditional classical electromagnetism ignores the mass
textbooks when there is no resistance in the circuit; if there contribution to the inertia that appears in Eq. (25) and
is no resistance, the self-induced emf cancels the external regards the self-inductance as a geometrical quantity associ-
emf and the total force per unit charge is zero. ated with the approximation given in Eq. (26). Thus,
Most physicists find it surprising that the net force on each the self-inductance L of the multi-charge ring system from
particle is zero and yet the particles are accelerating. In fact, Eq. (26) is
the zero-force result from Eq. (27) is only an approximation;
the full equation is given in Eq. (24) where indeed Newton’s 2pREe ðRÞ ð2pÞ R X 2
N 1
2  3 sin2 ð pj=N Þ
second law holds. There is a self-consistent relation between L¼ ¼ :
Neðd2 /=dt2 Þ=ð2pÞ 2c2 N j¼1 2 sinð pj=N Þ
the common acceleration of each charge and the back elec-
tric force on each charge due to the accelerations of the other (29)
charges. In the limit of a large number of charges, we can
compute the common acceleration simply by insisting that We see that the self-inductance of this multiparticle,
the acceleration field back at any charge due to the accelera- circular-orbit circuit is now independent of the mass m and
tion of the other charges should (approximately) balance the of the charge e of the current carriers.18 As with all expres-
external force per unit charge. The situation is perhaps easier sions for the self-inductance L of a circuit, Eq. (29) has
to grasp if we imagine multiplying Eq. (24) by the common dimensions of length divided by c2 . In the appendix, the self-
velocity of each particle. Then the equation states that the inductance L in Eq. (29) for our hypothetical circuit is

268 Am. J. Phys., Vol. 83, No. 3, March 2015 Timothy H. Boyer 268

This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: On: Tue, 24 Feb 2015 15:35:23
connected to the self-inductance of a circular wire of nonzero goes only slightly into particle kinetic energy but mainly is
cross section. converted into the magnetic energy associated with the
motion of the charges on the ring. This stored magnetic
energy is exactly ð1=2ÞLi2 .
4. Energy of the current carriers For our example of an external emf acting on a circular
The change in energy DU of the system associated with charged-particle circuit, the situation with many particles is
nonzero velocity v for the charges of the circuit includes totally transformed from the situation with only one particle.
both mechanical kinetic energy DUmech and magnetic field As charged particles of mass m and charge e are added to the
energy DUmag circuit, the mechanical inertia increases linearly with the
number of particles N while the inertia associated with the
DU ¼ DUmech þ DUmag mutual electromagnetic interactions increases quadratically
ð with N. In the multiparticle case, the mutual interaction
1 between the particles overwhelms the single-particle behav-
¼ mi c2 ðci  1Þ þ d3 r B2 : (30) ior so that the mass and charge of the individual charge car-
riers is no longer of significance. The particles move so that
The self-inductance L of the circuit is associated solely with the sum of the induced electric fields at each particle approx-
the magnetic energy stored in the DUmag ¼ ð1=2ÞLi2 . imately cancels the external electric field. The energy of the
Ð circuit,
3 2 charge carriers of the ring includes both mechanical kinetic
The magnetic energy Umag ¼ d r B =ð8pÞ stored in the
circular-orbit circuit is given by the cross terms (but not the energy and magnetic energy, but the magnetic energy is
self-terms) when the magnetic field is squared, and corre- dominant in the multiparticle limit. For the situation envi-
sponds to the velocity-dependent double sum in the Darwin sioned in the textbook discussions of Faraday induction, the
Lagrangian, Eq. (5). Thus we have mechanical energy is so small compared to the magnetic
energy that the mechanical energy is never mentioned.
ð We notice that in Eq. (8) if the resistance R of the circuit
1X N X
DUmag ¼ d 3 r 2Bei ðrÞ  Bej ðrÞ vanishes, then the current increases at a steady rate di=dt ¼
2 i¼1 j6¼i 8p emf ext =L without limit. Of course, Eq. (8) involves the self-
" # inductance L, which has no role for the inertia or kinetic
1X N X 2
e vi  vj vi  ðri  rj Þvj  ðri  rj Þ
energy of the charge carriers because these are assumed min-
¼ þ
2 i¼1 j6¼i 2c2 jri  rj j jri  rj j3 iscule compared to the electromagnetic mutual interactions
" # and magnetic energy. Since our analysis has used the low-
N 1 2
e v^y  vj y  ðrj Þ vj  ð x^RÞ
v^ velocity approximation of Eq. (3), our treatment assumes
¼N þ ; (31)
2c2 j^ x R  rj j x R  rj j3
j^ ðv=cÞ2  1 and cannot be extrapolated to velocities compa-
rable to the speed of light.
where in the last line of Eq. (31) we have used vi  ri ¼ 0 and
have taken advantage of the symmetry to evaluate the mag- V. DISCUSSION
netic energy when the Nth particle is located on the x-axis at
rN ¼ x^R and is moving with velocity vN ¼ y^v ¼ y^Rd/=dt. We have discussed the Faraday induction from an unfami-
The jth particle is located at rj ¼ R½^ x cosð2pj=NÞ þ liar particle point of view, including a detailed treatment of a
y^ sinð2pj=NÞ with velocity vj ¼ Rðd/=dtÞ½^ x sinð2pj=NÞ simple hypothetical circuit. We have evaluated the electric
þ^ y cosð2pj=NÞ. Introducing these expressions along with fields of individual electric charges and shown how a system
the distance given in Eq. (21), the magnetic energy of involving a single charge is transformed over to a familiar
Eq. (31) is electromagnetic system as the number of charges is
increased. Our example involves charges under centripetal
 2  constraining forces giving a circular orbit but allowing tan-
N 1 2
e d/ cosð 2pj=N Þ
DUmag ¼ N R gential acceleration. In the one-particle example, the induced
2c 2 dt j2sinð pj=N Þj
j¼1 electric field depends upon both the charge e and the mass m
# of the charge carrier with the induced field proportional to
sin2 ð 2pj=N Þ e2 =m. In the one-particle case, the induced electric field is
j2sinð pj=N Þj3 small, and the energy transferred to the particle goes into the
8 " #9 kinetic energy of the ring particle. The multiparticle circuit
1 <½2pÞ R X 2  3sin2 ð pj=N Þ = eN d/
2 N1
has a completely different behavior from that given by a
¼ : summation over many one-particle circuits with the one-
2 : 2c2 N j¼1 2sinð pj=N Þ ; 2p dt
particle acceleration. In the multiparticle case, the electro-
(32) magnetic forces between the charges are such as to transform
the behavior over to the familiar behavior of a conducting
We recognize the current i ¼ eNðd/=dtÞ=ð2pÞ and so circuit where the charge and mass of the current carriers are
can read off the self-inductance of the circuit from DUmag ¼ of no significance. When there are a large number of charged
ð1=2ÞLi2 in Eq. (32). The expression for the self-inductance particles, the acceleration of each charge is determined by
L is the same as in Eq. (29). Again for an interacting multi- essentially the requirement that the sum of the acceleration
particle system, the mechanical kinetic energy increases fields of all the other charges should cancel the external force
linearly with the number of particles N while the magnetic per unit charge that produces the external emf around the cir-
energy increases as N 2 . In this multiparticle limit, the power cuit. This crucial mutual interaction of the current carriers is
Pi ¼ efext vi delivered to the ith charge by the external force sometimes not appreciated in the traditional textbook
per unit charge f ext associated with the original external emf treatment.

269 Am. J. Phys., Vol. 83, No. 3, March 2015 Timothy H. Boyer 269

This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: On: Tue, 24 Feb 2015 15:35:23
ACKNOWLEDGMENTS Accordingly, the self-inductance of our hypothetical
circuit is
The author wishes to thank Dr. Hanno Essen for sending
copies of the work of C. G. Darwin and of his own work, listed     
ð2pÞ2 R N 2N 2N
now in Refs. 3 and 4. The author had been unaware of these L 2
contributions. Also, the author wishes to thank a referee for 2c N p p p
many helpful comments on earlier versions of this article, 4p 2N
which directed attention toward needed clarifications. ¼ 2 R ln 1 : (A4)
c p

The self-inductance of a circular loop of wire of radius R and

circular cross section of radius a is given by20
L ¼ 2 R ln  : (A5)
The self-inductance of our hypothetical circuit of N c a 4
charges moving in a circle of radius R was given in Eq. (29).
In order to compare this expression with that of a continuous The separation between the charges of our hypothetical cir-
wire, we approximate the sum by an integral. Using the tra- cuit is Ds ¼ 2pR=N. Evidently, if we take the effective
pezoid rule,19 radius a of the cross section of the hypothetical circuit as
" a  2Ds, then there is agreement for the logarithmic terms
ba X n1   between Eqs. (A4) and (A5). For large values of N, the loga-
dx f ðxÞ ¼ f a þ kðb  aÞ=n rithmic term should be the dominant contribution.
a n k¼1

ð Þ
þ f ðaÞ=2 þ f b =2 ; (A1) 1
See, for example, the undergraduate textbook by D. J. Griffiths,
Introduction to Electrodynamics, 3rd ed. (Prentice-Hall, Upper Saddle
River, NJ, 1999).
we have 2
See, for example, the graduate-level textbook by J. D. Jackson, Classical
Electrodynamics, 3rd ed. (Wiley, New York, 1999).
" # pp=N
ð " # 3
N1 C. G. Darwin, “The inertia of electrons in metals,” Proc. R. Soc. London
2  3 sin2 ð pj=N Þ N 2  3 sin2 ðxÞ A 154, 61–66 (1936).
¼ dx 4
2 sinð pj=N Þ p 2 sinð xÞ One of the rare instances is provided by the work of H. Essen, “From least
p=N action in electrodynamics to magnetomechanical energy—A review,” Eur.
" # J. Phys. 30, 515–539 (2009). In an appendix, Essen evaluates the self-
1 2  3 sin2 ðp=N Þ inductance of the same hypothetical circuit as used in the present article.
þ 5
Attention to a current-source point of view is encouraged by S. E. Hill,
2 2 sinðp=N Þ
" # “Rephrasing Faraday’s law,” Phys. Teach. 48, 410–412 (2010), who refers
1 2  3 sin2 ðp  p=N Þ to Jeffimenko’s equations when seeking the source of the Faraday induc-
þ : tion field. Hill’s reminder is mentioned in a footnote by D. J. Griffiths,
2 2 sinðp  p=N Þ Introduction to Electrodynamics, 4th ed. (Pearson, New York, 2013),
p. 313.
(A2) 6
See, for example, Ref. 2, p. 664. When treating relativistic aspects of elec-
tromagnetism, Gaussian units are far more natural than S.I. In S.I. units,
Ð integral can be evaluated analytically using one would have a factor of 1=ð4peo Þ multiplying the right-hand side of Eq.
dx= sin x ¼ ð1=2Þ½lnð1  cos xÞ  lnð1 þ cos xÞ. Then not- (1) and a factor of 1=c multiplying the right-hand side of Eq. (2).
ing sinðp  p=NÞ ¼ sin p=N and cosðp  p=NÞ ¼ cos p=N, 7
L. Page and N. I. Adams, Electrodynamics (D. Van Nostrand, New York,
and expanding in powers of 1=N, the sum on the first line of 1940), p. 175. This text was reprinted by Dover Publications, New York,
Eq. (A2) becomes in 1965, but now seems to be out of print.
" # C. G. Darwin, “The dynamical motions of charged particles,” Philos. Mag.
N 1
2  3 sin2 ð pj=N Þ 39, 537–551 (1920). The Darwin Lagrangian appears in some modern
textbooks. See, for example, Ref. 2, pp. 596–598, or L. D. Landau and E.
2 sinð pj=N Þ M. Lifshitz, “The classical theory of fields,” 4th ed. (Pergamon, New
  York, 1975), pp. 165–168.
N 1 1  cosðp  p=N Þ 1 þ cosðp=N Þ 9
See, for example, Ref. 1, pp. 458–459.
¼ ln 10
p 2 1 þ cosðp  p=N Þ 1  cosðp=N Þ L. Page and N. I. Adams, “Action and reaction between moving charges,”
    Am. J. Phys. 13, 141–147 (1945).
3 p 3 p 11
T. H. Boyer, “Lorentz-transformation properties of energy and momentum
þ cos p   cos
2 N 2 N in electromagnetic systems,” Am. J. Phys. 53, 167–171 (1985).
" # " # 12
T. H. Boyer, “Example of mass-energy relation: Classical hydrogen atom
1 2  3 sin ðp=N Þ 1 2  3 sin2 ðp  p=N Þ accelerated or supported in a gravitational field,” Am. J. Phys. 66,
þ þ 872–876 (1998).
2 2 sinðp=N Þ 2 2 sinðp  p=N Þ 13
(  See, for example, Ref. 1, p. 303.
  ) 14
See, for example, Ref. 2, problem 14.24 on pp. 705–706. There is no radia-
N 1 þ cosðp=N Þ p
¼ ln  3 cos tion emitted, and the fields of the circuit are those of electrostatics and
p 1  cosðp=N Þ N magnetostatics.
" # 15
Essen in his appendix discusses the necessary approximations to connect
2  3 sin2 ðp=N Þ the self-inductance of this hypothetical circuit back to that of a wire of
þ nonzero thickness that is bent in a circle. See the appendix of Ref. 4. We
2 sinðp=N Þ have followed Essen’s basic analysis in the appendix of the present article.
N 2N 2N See, for example, Ref. 1, pp. 271–272.
¼ 2ln  þ Oð1=N Þ: (A3) 17
The situation of our one-particle example contains elements of the same
p p p physical system as “the Feynman disk paradox.” See R. P. Feynman, R. B.

270 Am. J. Phys., Vol. 83, No. 3, March 2015 Timothy H. Boyer 270

This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: On: Tue, 24 Feb 2015 15:35:23
Leighton, and M. Sands, The Feynman Lectures on Physics (Addison- N ¼ 3. Only for N ¼ 4 does the summation first become positive ðþ0:207Þ.
Wesley, Reading, MA, 1964), Vol. II, pp. 17-5 and 27-6. See also Ref. 1, It should be emphasized that even for small numbers of charges, provided
pp. 359–361. In the present article, we evaluate the induced electromag- that we observe the restriction that mR  e2 =c2 , the total inertia is posi-
netic field arising from the accelerating disk charges by use of the approxi- tive. For a few charges, the mass of the current carriers is crucial; for large
mate electric field expression (3). N, the mass of the current carriers is unimportant. The negative value of L
A continuous circular wire of nonzero cross section provides the analogue for a small number of charges corresponds to total energy in the combined
within traditional electromagnetism texts of our hypothetical circuit. Our magnetic fields that is smaller than that of the individual magnetic fields of
hypothetical circuit is not continuous but rather involves a finite number N the charges when located far from each other. The Darwin–Lagrangian
of charges with spaces between the charges. It seems interesting that for approximation contains some of the run-away aspects that are seen else-
our discrete circular charge arrangement, the typical multiparticle behavior where in classical electromagnetic theory.
requires at least four charges. The summation in Eq. (29) is steadily See, for example, M. D. Greenberg, Advanced Engineering Mathematics
increasing with increasing particle number N. However, the summation (Prentice Hall, Upper Saddle River, NJ, 1998), p. 312.
starts out negative ð0:5Þ for N ¼ 2, and is still negative ð0:289Þ for See, for example, Ref. 2, p. 234.

271 Am. J. Phys., Vol. 83, No. 3, March 2015 Timothy H. Boyer 271

This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: On: Tue, 24 Feb 2015 15:35:23