You are on page 1of 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/230902660

About Poynting's theorem

Article  in  European Journal of Physics · July 2000


DOI: 10.1088/0143-0807/13/3/003

CITATIONS READS

18 1,730

2 authors:

Ignacio Campos j. l. Jimenez


Universidad Nacional Autónoma de México Universidad Nacional Autónoma de México
41 PUBLICATIONS   188 CITATIONS    74 PUBLICATIONS   382 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

this papaer is not mine View project

All content following this page was uploaded by j. l. Jimenez on 22 March 2016.

The user has requested enhancement of the downloaded file.


Home Search Collections Journals About Contact us My IOPscience

About Poynting's theorem

This content has been downloaded from IOPscience. Please scroll down to see the full text.

1992 Eur. J. Phys. 13 117

(http://iopscience.iop.org/0143-0807/13/3/003)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 144.122.220.41
This content was downloaded on 13/03/2016 at 22:14

Please note that terms and conditions apply.


I About Poynting's theorem

I Campos and J L Jimenez

Abstract. We review the general question of the physical Resumen. Se hace una revision del problema de la
interpretation ofPoynting's theorem and we compare two interpretacion fisica del tmrema de Poynting y se
typical derivations of it. One of these uses the work done comparan dos tipicas deducciones de CI. Una de e m s usa
on a charge by an exrernol electromagnet field, while the el trabajo hecho sabre una c a r s por el campo
other considers the work done by the roral field, external electromagnetico exlerno, mientras que la otra considera
plus ser/-field; though this derivation is mathematically el trabajo hecho par el campo rorol, que e6 el extern0 mas
consistent; it leads to difficult problems of interpretation. el auto-campa Aunqge esta deducci6n es
matematicamente consistente, conduce a diliciles
problemas de interpretacih

1. lntroductlon energy is taken seriously in stationary situations, as


when we have a point charge and a magnet at rest.
The principle of conservation of energy is one of the Much work has been done is the assignment of
most general and important in physics. Its general- energy, linear and angular momentum to the
ization to include electromagnetic fields and charges electromagnetic field as a physical system. In general
in interaction leads to Poynting's theorem: it seem that the ascription to the electromagnetic field
of these mechanical attributes can he done in a way
that is consistent with the usual interpretation of
Poynting's vector even in static situations (Feynman
+
where U = f (%E* B*/&) is the electromagnetic e l a1 1963, Romer 1966, 1967, Pugh and Pugh 1967,
energy density, E and B being the electric and magnetic Pugh 1971, Corinaldesi 1980, Lambardi 1983, Sharma
fields; S = (E x B)/& is the so-called Poynting vector 1988). However, the arguments used relate changes in
and j is the current density. static global configurations of charges and fields, thus
It is usual to interpret S a s an 'energy current' or as again leaving room for perplexity when the local flux
kGX 'oui ibis inierpreiaiion is fuii of of miigy is takin as a iiiizpoi: of energy cien i i i
difficulties. stationary situations. Therefore the controversy persits
In the first place it is easily seen that as a local flux even today when S is interpreted as a local flux of
of energy the vector S is not uniquely defined. In fact energy.
we could add to S any arbitrary divergenceless vector Although much less discussed in the literature,
and still satisfy equation (I). Thus some alternative another controversial point refers to the interpretation
definitions for the local flux of energy have been of the right-hand side of equation ( I ) . What is the
proposed (Slepian 1942, Lai 1981, Peters, 1982, meaning of E? Is it the external field, i.e. the field
Romer 1982, Kobe 1982). related to charges and currents outside the region of
Besides, when S is used to calculate the electromag- interest, or is it the roral field, the external field plus
netic momentum of a classical model of the electron, the field produced by charges and currents inside our
+
a factor appears in the so-called electromagnetic region?
These are not mere linguistic difficulties and we find
mass and this is usually regarded as being inconsistent
with the special theory of relativity (Campos and old and recent texts that caution us that Poynting's
Jimenez 1986). vector 'cannot, except in special cases, he identified
These are not the only problems with Sinterpreted with the energy transport vector' (Wilson 1940) and
as a local energy flux, with some people disagreeing that 'It may not yield the correct energy flux at every
even about the direction of the energy flow (Gough given point' (Hauser 1971).
1982). Besides some well known physicists-like Since a revision of the derivations could he a good
Feynman (Feynman er a1 1963), Eyges (1972) and way to clarify this issue, we analyse two derivations of
Beketi (Bekefi and Barret 1977wonfess their perplex- Poynting's theorem that difkr in their interpretations
ity when the interpretation of S as a local flux of of the field E. We will show that although they form-
118 I CamDos and J L Jimbnez

ally lead to the same continuity equation, these in our region the total fields E = E ' + E h and
derivations are not equivalent. B = B' + Bh, where i and h stand respectively for
inhomogeneous and homogeneous.
An important question is the relationship that these
2. Some general remarks on classical fields hear to the external field and to the se(ffield, a
electrodynamlcs distinction that it is necessary lo make for the inter-
pretation of the action of the field on a point charge.
As a field theory, classical (microscopic) electro- Since the external field is not singular at the location
dynamics constitutes a hyperbolic system of partial of a (finite) point charge, it must be a homogeneous
differential equations. These equations describe the solution. However, the definition of the self-field is a
evolution of the electromagnetic field if the charge- delicate problem that involves additional postulates
current distribution is given. and cannot he done in general (Bergman 1962). This
In other words, the field equations '. , , determine is an important theoretical problem that must he
the development of the field in the course of time if the solved 6efore appealing to any measurement. As
initial field as well as the distribution of charges at all Einstein once explained to Heisenberg (Heisenherg
times are known' (Bergman 1962). On the other hand, 1990), 'whether you can observe a thing or not
the ponderomotive effect of the electromagnetic field depends on the theory you use; it is the theory which
on test point charges is given bv the Lorentz force law. decides what can be observed'. Since the self-field is
+
F = q(h U x B). Thisforce iaw .determines the
I':. singular at the location of a finite point charge, it must
acceleration of a particle in the presence of a given he the inhomogeneous field or part of it. Usually it is
'incident field', (the [finite]diKerence between the total taken as the velocity-dependent inhomogeneous field,
field and the 'self-field')"' (Bergman 1962). The also called the convected or hound field (Wilson
incident or external field has no singularity at the 1936), in contrast with the acceleration-dependent
location of the particle on which it acts, while in the part (identified as a eadiation field) which eventually
case of a (finite) point charge the self-field is infinite at becomes detached from the charge.
the location of the charge. Some authors (Goedecke 1975) take the self-fieldas
The separation of the total field into an incident or the whole inhomogeneous field, but in the case of a
external field and a self-field is related to the 'source charge moving with constant velocity this ambiguity
hypothesis', that is, to the assumption that the does not appear.
charged bodies are the sources of the electromagnetic The problem of disentangling the self-field from the
field. However, ' . . . the source hypothesis is an extra total field has been treated in several ways, the best
assumption: as our axiom system shows, fields and known being that proposed by Dirac (1938) and later
currents are conjoined but not causally associated: extended by Teitelboim el a1 (1980). Essentially this
only field changes are causally associated with consists in defining the self-field on a spacelike hyper-
charged bodies, in case there are any in the region surface orthogonal to the worldline of the particle. In
considered' (Bunge 1967). this manner the infinite energy associated with the
Classical electrodynamics considers then two types self-field can he renormalized covariantly. In spite of
of entities, charged particles and fields, but their its advantages, this treatment was meant by Dirac
interaction can only be oartiallv analvsed. since either only as a phenomenological theory, useful to calculate
thedistnbution ofiharies andcurreit is known and dl1 the result( that Cdn be ohtained from expcnment.
then we can denve the evolution of the field. or an but inthi,own%ordi, he wasnevercompletely sitiqfied
e x t m " field acting o n thc particles is given and then with this 'subtraction physics'.
we can find the motion of the particles. This dualistic Besides the problem of defining unambiguously the
nature of classical electrodynamics has provoked such self-field, one has also the problem of determining the
a discomfort since long ago (Pauli 1Y73). that alterna- action of this field on the charged particle. Usually
tive theories have been proposed (Einstein 1976. this action is taken as given by the same Lorentz law
Hoyle and Narlikar 1974). hut this is an additional postulate that may be wrong
The interation of charges and fields is then (Bunge 1967). In fact, the problem of the interaction
described by the (microscopic) Maxwell equations of a charged particle with its self-field leads to well
and the Lorentz force law. If the trajectories of the known difficulties in the characterization of the
charges in the region of interest are known, then what radiation reaction (Jimenez and Campos 1987).
rcmams IS only the mathematical problem of solving With these remarks in mind, we now analyse two
an inhomogeneous system of differential equation,. different derivations of Poynting's theorem and the
As is well known, the general solution is the sum of problems of interpretation posed by them.
the general solution of !he homogenous system (uith
all the charges equated to zero) plus a particular
solution of the inhomogeneous system. 3. Derlvations 01 Poyntlng's theorem
This is not all. since a unique and stable solution
further requires the imposition of initial and boundary The hest known derivation appears in Jackson's text
iUlidlilUlij Gil ihr IreiGj. w e ira,e iiicn ai i . \ C i j pxii: :!9?5:. U- k g i x byca-!ca!aticg !he p w e r !ransfxed
About Poynting's theorem 119

to a charge by an external field through the Lorentz is defined as


force:
1
JW S=-(EhxBh) (9)
-=
Ji q(Eh + w x B h ) . w = p .~h k
and one finally gets
where we have used the superscript h to recall that
the extemal field is a solution of the homogeneous - j . E h = V . A " + za.d
equations. (10)
This energy is transferred from the external field to
the particle and it may appear as kinetic energy
or other forms of energy. This is consistent with As we have shown, Jackson's derivation of Poynt-
ing's theorem is not completely clean. The inter-
Newtonian mechanics that demands that only exter-
nal forces may change the energy of a particle. pretation ofequation (IO) as an energy balance would
require that the energy of the external field diminishes
The next step is to eliminate the current j = qn from
the last equation. We take it from the Ampere- on acting on a charge and giving rise to the current
j = qw. This would be the case, for example, for an
Maxwell law,
electron accelerating in the TV tube. However, the
dE' energy balance (IO) does not take into account the
V x P=p++-++j. (3) possible transference of energy to the self-field of
ai
the charge.
Here the complete fields appear, since they are the The second type of denvation of Poynting's theorem
general so!utions of the inhomogeneous cquatioxs. is based only on Maxwell's equations, and involves
Hence, in substituting j from the Ampere-Maxwell therefore the work done on a charge by rhe roialfields.
law, we introduce the self-field into the derivation. We are going lo write these fields without any
Then we have superscripts.
Formally this derivation is logically consistent, but
the presence of self-fieldsbrings complex problems of
interpretation.
This equation can be transformed using the vector The derivation is a formal transformation of
identity Maxwell's equations (Good and Nelson 1971). It
V . ( E hx E ) = E . V x Eh- E h . V x B' ( 5 ) begins by multiplying by E the Ampere-Maxwell law:

and Faraday's law, obtaining

JE
=j . E + E ~ E-.
7%

Using again the identity for the divergence of a


cross-prodoc! and Filrilday's !ay, 2nd fo!!owing thc
This mixture of homogeneous and complete fields same steps as before, we obtain
prevents the simplification usually done when this
distinction is neglected. In order to arrive at the final V . (E x %)+ ;(&BZ + * + E ') = -j. E. (12)
result, it is necessary to make a risky move: to discard
the inhomogeneous part of the complete fields. By Identifying S a n d U as the Poynting vector and the
doing this one obtains energy density for the total fields, we obtain Poynting's
theorem, equation ( I ) . Though seemingly the same
continuity equation results, it is important to note
that equations (7) and (12) are different, the latter
involving the total fields, while the former refers only
to the external fields.
Since in this derivation the total fields have been
The last two terms on the right-hand side can be introduced, we need to interpret the term -j E = .
written as - j ' (E" + E""). Usually this term is interpreted as
the power developed by the total fields. This amounts
to postulating that in the Lorentz force law it is
J l I
--(-(Bh)'
ai2
+ +(Eh)') =ad
T (8) legitimate to put in the total fields, not just the external
fields, but this would be a daring assertion. Experience
defining in this way an energy density uh for the guarantees the Lorentz force only for external fields
homogeneous or extemal fields. (Bunge 1967) and the introduction of the total fields
The corresponding Poynting vector for these fields raises several questions. Can the self-fieldsdo work on
120 I Campos and J L Jimenez

the charge? An affirmative answer would mean that a Poynting's theorem as the local energy conservation
charge can accelerate itself through the action of its law and Poynting's vector as a local transport of
self-field, in contradiction with experience. On the energy, has led us to analyse two types of derivation
other hand, if the self-field can do work on a of this theorem. The derivation based on the work
charge only in the presence of an external field, what done on a charge by an external field is not satisfactory
fraction of the work is attributable to the self-field because it introduces a mixture of external and total
and how much to the external field? These are open fields. An interpretation of this version of the theorem
questions that present-day electrodynamics is far as a transfer of energy from the external field to the
from answering. charge is possible only if the self-field is ignored, as
With regard to the meaning o f j in the term E . j , we implicitly done in the textbooks.
have implied that it is as if j = p, thus remaining The other version of the theorem amounts to a
within the scope of microscopic electrodynamics. As formal transformation of Maxwell's equations. It is
mentioned in the introduction, we still have the logically consistent but by introducing the self-field
problem that the electromagnetic momentum derived poses challenging problems that compel us to look
from Poynting's vector and the electrostatic self- beyond present-day electrodynamics.
energy of a point charge, do not form a 4-vector. This Then the physical meaning of Poynting's theorem
is because the 4-divergence of the electromagnetic would be clear cut only in regions free of charges,
stress-energy tensor is not zero. giving a satisfactory picture in the case of a plane
Two approaches have been proposed lo overcome wave. However, the theorem should be used with care
this problem. One consists in introducing Poincart in the presence of charges. Besides we may add that
stresses to have a truly closed system, thus imposing any prescribed current, microscopic or macroscopic,
formally the conversation of toro1 energy-momentum, requires the intervention of non-electromagnetic
i.e. electromagnetic plus non-electromagnetic. The forces. Then we may conclude that a system of
other point of view proposed the redefinition of electromagnetic fields and charges is usually not a
the electromagnetic momentum of a point charge, closed system. More than that, in the case of the
modifying the momentum given by Poynling's vector. electron nobody knows which fields may close thc
In fact it has been argued (Rohrlich 1970) that system. As an alternative approach, Dirac's theory
Poynting's vector defines correctly the electromagnetic treats the electron as a true singularity. Other alterna-
momentum only for zero rest mass systems, i.e. radi- tives include non-linear field theories (Einstein 1976)
ation, while the presence of a point charge with rest and relativistic action-at-a-distance theories (Wheeler
mass greater than zero requires a different definition and Feynman 1949, Hoyle and Narkilar 1974). All
such that the rest energy and the redefined momentum these represent valuable efforts in the long struggle to
consitute a 4-vector. understand the structure of matter.
The case of a conduction current in the term E .j,
as used in many texts to illustrate an application of
Poynting's theorem, would take us into the domain of Acknowledgements
macrosenpic electrodynamics. The analysis of such an
application deserves a separate and careful treatment, We want to express our gratitude to Professor D
and we just mention the existence of a conduction Moreno for his critical comments on this work and to
current requires a non-electromagnetic source of Universidad Autonoma Metropolitana, Iztapalapa,
energy. This is apparent from the need to introduce a for its hospitality
+
generalization of Ohm's law: j = u(E Fe"),where
F'" is an electromotive force of non-electromagnetic
nature (mechanical, chemical, etc) that produces a References
known motion of charges, the so-called 'external
current'y =OF". Bergman P G 1962 Basic Theories of Physics (New York:
In this way the Ampere-Maxwell law becomes Dover) pp 266, 177, 196
Lkkefi G and Barrel A H 1977 Elecrromognelic Vibraions,
B aE W o w and Rodiarion (Cambridge, MI:MIT Press) pp
V x -=E+-
~b ar
+uE+y 236-8
~~ ~

Bunge M 1967 Foundorions o/Physics (Berlin: Springer)


This equation yields the telegrapher equation pp 173, 165, 174-5
instead of the usual wave equation. As we can see, a Campos I and Jimenez 1 L 1986 Phys. Rev. D 33 607-10
cogent application of the energy conservation law Corinaldesi E 1980 Am. J . Phys. 40 8 3 4
requires the consideration of a closed system that Dirac P A M 1938 Proe. R . Soc. A 167 148-69
Einstein A 1976 The Meaning of Reloiiviry (London:
includes the electromagnetic system as part of it. Chapman and Hal!) p I57
Eyges L 1972 The CIpssical Eleerromognelic Field
(Reading: Addison-Wesley)pp 200-3
4. Concluding remarks Feynman R P Lcighton 8 and Sands M 1963 The
Feynman Leerures on Physics Val 2 (Reading. Addison-
The '--i!deT.-en: :ha: coxies from int.rpreti!!g Wes!ey) pp 2 7 4 !U 27-8, !?A
About Poynting's theorem 121

Goedecke G H 1975 Nuovo Cimenfo B 28 22543 (Cambridge, MI: MIT Press) pp 2-3
Good R H and Nelson T J 1971 Clarsieol Theory of Peten P C 1982 Am. J . Phys. 50 1165-6
Eleerric and Magnetic Fie/& m e w York: Academic) Pugh E M and Pugh G F 1967 Am. J. Phys. 35 153-6
PP 264-5 Pugh E M 1971 Am. J . P h p . 39 837-8
w
G O U ~ ~1982 E,,,.J. m y S . 3 83-7 Rahrlich F 1970 Am. J . Phys. 38 1310-6
Hauser W 1971 Introduction l o the Principles of Romer R H 1966 Am. J. Phys. 34 772-8
Electromaewrirm (Reading: Addison-Weslev) I) 399 -1967 Am. J. Phys. 35 44-6
Heisenberg % 1990 b l f r c a f t & of Fundnmenl~l~orces
ed -1982 Am. J. Phys. 50 11668
A Salam (Cambridge: Cambridge University Press) p 99 Sharma N L 1988 Am. J . Phys. 56 420-3
Hoyle F and Narlikar J V 1974 Acfion ai a Distance in Slepian J 1942 J . Appl. Phys. 13 512-8
_ielteibaim
.
Physics and Cosmology (San Francisco: Freemanj C Viiiarroei D and van weert Ch 0 i9uu Rip.
Jackson J D 1975 Clossicol Electrodynamics 2nd edn Nuovo Cimenro 3 (9) 1-63
(New York Wiley) pp 236-7 Wheeler J A and Feynman R P 1949 Rev. Mod. Phys. 21
Jimbnez I L and Campos I 1987 Am. J . Phys. 55 1017-23 425-33
Kobe D H 1982 Am. J . Phys. M 1162-3 Wilson W 1936 Proc. Phys. Soc. 48 736-40
hi C S 1981 Am. J . Phys. 49 841-3 -1940 Theoretical Physics Vol 3 (London: Methuen)
Lombardi G G 1983 Am. J . Phys. 51 2 1 3 4 P 53
Pauli W 1973 Pauli Lectures on Physics Vol I

View publication stats

You might also like