Академический Документы
Профессиональный Документы
Культура Документы
Cheyenne J. Sheppard
ABSTRACT
Cheyenne J. Sheppard
Optical momentum within material media has been studied for a century. Although great strides
have been made towards understanding photon momenta, there is still a question regarding the
correct momentum of light. This is due to the many mathematical interpretations of Maxwells
tities. Within this correspondence, the optical momentum controversy is studied via mathematical
The electromagnetic wave momentum is studied via moving contribution, where a framework
for moving media is developed. Application of the framework with respect to independent formula-
tions of electrodynamics are shown to be consistent with conservation of energy and momentum. In
doing this, thought experiments are employed to test each approach and demonstrate the electrody-
namic forces with respect to each formulation. This is first applied to two moving perfect reflector
submerged in a dielectric medium such that both Abraham and Minkowski momenta are derived.
Second, a moving slab of moving magneto-dielectric material is studied expanding the arguments into
time average and time varying calculations. Each approach is consistent with electromagnetic wave
theory and demonstrates the mathematical methods for deriving the both Abraham and Minkowski
momentum models.
The kinetic subsystem is studied by use of established physical principles, namely the Lagrangian.
The field energy density and momentum density are used to derive the force expressions of electro-
magnetic fields. This formulates the respective system of equations, forming Maxwells equations
in terms of the field energy and momentum density. By use of the relativistic principle of virtual
power, the Maxwell field stress tensor and kinetic momentum density are derived uniquely, disproving
ii
.
ACKNOWLEDGMENT
I would like to acknowledge and sincerely thank my research advisor, Dr. Brandon A. Kemp.
During the course of my research, I have had the great pleasure of learning invaluable skills and
techniques for modeling physical systems from Dr. Kemp. I have also learned a lot about electro-
magnetic wave theory and dealing with fundamental electromagnetic concepts, which extend well
into other predominate areas of physics. With this, I acknowledge Dr. Kemps depth in understand-
ing and enate ability to explain and illustrate difficult concepts and ideas. I extend my deepest
gratitude to Dr. Kemp for the knowledge and opportunities that he has provided me.
I would like to extend a special thanks to Dr. Jie Miao, my academic advisor Dr. Debra Ingram,
and the Department of Mathematics and Statistics at Arkansas State University. This thesis would
not have possible without the support and motivation from faculty and staff within this department,
I would also like to thank my family. Throughout my academic pursuits, my parents, Robert and
Lorie Sheppard, have given me unconditional support and have urged me to pursue my academic
goals. Additionally, my sister, Rachael Sheppard, has shown great passion and support towards my
academic endeavors. Finally, I would like to thank my fiancee, Ashley Jackson, for her unwavering
iii
.
Contents
1 Introduction 1
1.2.1 Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Mathematical Framework 9
3.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
iv
4.1.1 Hamiltons variational principle . . . . . . . . . . . . . . . . . . . . . . . . . . 35
5.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
6 Conclusion 69
6.4 Addition of the material subsystem for time varying moving media . . . . . . . . . . 72
Appendices 73
B Field Relations 77
v
References 87
vi
List of Figures
p
1 A plane wave propagating within a dielectric with refractive index n = /0 nor-
mally incident onto a perfect reflector, with reflection coefficient Rmirror = ei . The
2 The mechanical force and work versus velocity for both the Minkowski and Chu
formulations are presented for the PEC. Here the normalizing factors are hSi i =
E0 H 0 v
2 and = c is the normalized velocity, where n = 3.25. The Minkowski and
Chu momenta are equivalent and tend to infinity when the relation n = 1. The
3 The mechanical force and work versus velocity for both the Minkowski and Chu
E0 H0
formulations are presented for the PMC. Here the normalizing factors are hSi i = 2
v
and = c is the normalized velocity, where n = 3.25. The Minkowski and Chu
5 The electromagnetic force and work versus velocity for all formulations are presented
vii
List of Tables
media. Here, the Minkowski formulation models the canonical force, which corre-
3 The derived values for the Minkowski subsystem. The Minkowski stress tensor and
momentum density are expressed in (a) and the Minkowski power flux and energy
4 The derived values for the time domain Minkowski subsystem. The Minkowski stress
tensor and momentum density are expressed in (a) and the Minkowski power flux and
5 The derived values for the time domain Chu subsystem. The Chu stress tensor and
momentum density are expressed in (a) and the Chu power flux and energy density
6 The derived values for the time domain Amperian subsystem. The Amperian stress
tensor and momentum density are expressed in (a) and the Amperian power flux and
viii
.
1 Introduction
In the mid 1800s, J.C. Maxwell formulated electromagnetic theory [2] unifying electricity, mag-
netism, and optics. This lead to the development of electromagnetic wave theory and Maxwells
prediction of radiation pressure. At the turn of the century, both Poynting and Einstein inde-
pendently validated Maxwells electromagnetic theory by the development of the special theory of
relativity (Einstein)[3] and detailed experimental radiation pressure calculations (Poynting)[4]. The
latter drew interest into optical forces and how light interacts with materials. In light of this, H.
of Maxwells electromagnetic theory [5]. Shortly after, M. Abraham suggested an alternate approach,
of which, utilized a more symmetric SEM tensor, yielding related, yet different quantities [6]. For
Minkowskis approach, the resulting electromagnetic momentum expression from the SEM tensor
included contributions from the material properties. However, Abrahams resulting electromagnetic
momentum within the SEM tensor renders a material-less value for the momentum expression. Be-
ing that both are mathematically valid, a debate arose as to which formulation accurately describes
Within the past decade, new discoveries and technologies have lead to a renewed interest in
optical momentum. Throughout the scientific community, optical forces have been, and are currently
used to manipulate and control matter [711]. With this, it is paramount to have a thorough
understanding of how light interacts with materials. To illustrate the enate discrepancies, we review
= nE,
B
pM = dV D (1)
c
1
the speed of light in vacuum. Conversely, the Abraham momentum,
= 1 E,
H
pA = dV 0 0 E (2)
nc
is the
expresses a decrease in electromagnetic momentum as light traverses a material, where E
is the magnetic field, 0 is the permittivity of free space, and 0 is the permutabil-
electric field, H
ity of free space. Due to classical electrodynamic relations, the Minkowski momentum density,
M = D
G B,
includes the material response within the integration such that D = 0 E
+ P and
According to the Minkowski momentum, this addition of material responses within the momentum
density gives rise to an increase in optical momentum as light propagates through a material. How-
A = 0 0 E
ever, the Abraham momentum density, G H,
renders a reduced momentum value due
to calculations including free space quantities. Recent advances have yielded a correlation identify-
ing the Abraham momentum as the electromagnetic field or kinetic momentum and the Minkowski
momentum is the canonical momentum of a given system [12]. Here, the kinetic momentum is re-
sponsible for the center-of-mass translation of a material while the canonical momentum describes
translations within or with respect to another material. Additional results rendered both the kinetic
sum to the same value when including the material contributions for each representation, where pkin
mat
and pcan
mat are the respective kinetic and canonical material contributions. In light of this, however,
many physical experiments render the Minkowski momentum as the observed photon momentum
[13, 14].
Alternative approaches have utilized the framework of quantum mechanics, recasting each mo-
E
pM = ~k = n (4a)
c
1E
pA = m
v = , (4b)
nc
where ~ is the reduced Planks constant, k is the wavevector, v is the velocity, and m is the photon
energy-mass equivalence E/c2 . With this, the quantum approach describes the Abraham momentum
2
Formulation force density (N/m3 )
Minkowski 2
12 E 2 + E
12 H + J B
Chu [ ( P )]E
( 0 M )H + J + P M
t 0 H 0 t 0 E
Einstein-Laub [ + (P )]E + (0 M )H + J + P 0 H
t
0 M 0 E
t
1 2 1 2 + J B +
Abraham 2 E 2 H + E t GM GA
P
Amperian ( P )E + t B + ( M ) B + E + J B
Table 1: Leading electromagnetic formulations and associated force densities applied to linear media.
Here, the Minkowski formulation models the canonical force, which corresponds to the Minkowski
momentum. The remaining formulations attempt to model the kinetic force, corresponding to the
Abraham momentum.
as the particle like momentum, while the Minkowski momentum is the wave like momentum. Al-
though this perspective is pragmatic and widely used [15], this representation disagrees with classical
electromagnetic theory [1]. This is due to both Abraham and Minkowski momenta being accurately
expressed by specific formulations of electrodynamics. Here, Table 1 demonstrates the various lead-
ing formulations commonly used within the literature [16]. With this, the Minkowski formulation
renders the Minkowski momentum, where the remaining formulations (Chu, Einstein-Laub, etc...)
render the Abraham momentum. Thus, each respective formulation of electromagnetic wave theory
The reason for this comes from each formulation being mathematically consistent with Maxwells
equations. That is, when summing each formulations field and material contributions, Maxwells
equations are obtained. However, depending on how each formulation describes the pure electro-
magnetic fields, the independent formulations render differing electromagnetic forces and thereby
physical interpretations. This allows for a myriad of differing physical models, all of which are con-
sistent with electromagnetic wave theory. For example, the Chu formulation defines the pure electric
and H,
and magnetic field vectors in terms of E respectively [17]. Thus, when modeling a system
with the Chu formulation, a material is conceptualized as containing electric and magnetic dipoles,
which interact with each respective field. Conversely, the Amperian formulation defines vector fields
and B
E as the electric and magnetic vector fields, respectively, rendering an electric dipole and
current loop for the associated physical model [18]. Although this is significant in conceptual vi-
sualization, it holds little importance for strict mathematical modeling the physical system. This
is due to the material parameters, or constitutive relations [18], of a given system being prescribed
and equally represented for each formulation. It is useful, however, to keep each physical model
in mind when mathematically expressing each physical system. This allows for one to utilize both
3
mathematical and physical reasonings when differentiating between the modeled relations, thereby
allowing for the understanding of the total dynamics of a system, and how they relate physically.
G
f = T (5)
t
being that we have predefined values for both the kinetic and canonical momenta in terms of the
momentum density, as presented in Eqs.(1) and (2)[5, 6]. Due to the relative insignificance and
trivial nature of the canonical system, the interest resides in which kinetic formulation accurately
models the kinetic system of electrodynamics. This is due to the canonical system being described
by the Minkowski formulation, where multiple formulations attempt to model the kinetic system.
When modeling the kinetic system, however, only one of the three values in Eq.(5) is known. That
= 0 0 E
is, the Abraham momentum density, G H,
is the only value known to satisfy the given
kinetic expression [19]. This allows for manipulations of Eq.(5) which lead to arbitrary force and
power expressions which satisfy the specified relation, but lend nothing to the understanding of the
physical system. Thus, the remaining question is which formulation renders the correct kinetic force
density and stress tensor to describe the true field relations inside materials? In this thesis, we look
to provide insight into the kinetic subsystem of light and demonstrate the optical momentum debate
within materials.
Over the past century, many researchers have made great strides into understanding optical mo-
mentum within media. Here, we review an abbreviated list, which allows for theoretical discussion
Radiation pressure on submerged mirrors. In the 1950s, Jones and Richards measured the
displacement of a submerged mirror due to incident radiation pressure within varying refracting flu-
ids [20]. Jones and Leslie reported in 1978 the use of a laser to produce similar experimental results
with higher accuracy [21]. In both experiments, the main purpose was to measure the deflection of a
small mirror suspended on a torsion balance while being subjected to optical pressures. Once these
pressures were observed in air, refracting liquids were introduced as a comparison to the vacuum
like case. The conclusion of the JRL (Jones-Richards-Leslie) experiments showed that the optical
4
pressure on a submerged mirror is proportional to the refractive index, n, of the submerging fluid.
Theoretically modeling the JRL experiments. Theoretical conclusions to the JRL exper-
iments have shown correlations between the optical momentum within a dielectric fluid along with
the type of reflector used, thereby rendering a specific momentum via the selection of reflector [22].
In theoretical experiments, one may choose to model a perfect electrical conductor (PEC) or a per-
fect magnetic conductor (PMC). Each conductor has specifically modeled parameter, such that at
the interface the PEC and PMC force either the electric and magnetic fields to zero, respectively. In
considering only dielectric materials, this allows for one to always derive the Minkowski momentum
Optical forces from moving boundaries. The analysis of a perfect conductor and dielectric
half-space resulted in a difference between the electromagnetic energy from of the incident and the
reflected and transmitted waves [23]. This energy difference was attributed to the media moving
with some velocity v perpendicular to a vacuum interface such that there is a net flow of electro-
magnetic energy. This phenomenon accounts for the change in the stored energy of the system, as
well as the work done by the mechanical forces put into the system. By use of energy-momentum
conservation laws, the calculated electromagnetic energy and pressures are rendered in terms of the
Balazss thought experiment. Previous research by Balazs [19] presents a simple box thought
experiment to determine the kinetic momentum of a photon by center of mass translation and con-
servation principles. Within this study, a photon with initial free space momentum ~/c is incident
upon a slab of material with refractive index n = . As the photon propagates within the ma-
terial, it is slowed with respect to free space propagation by length L = (n 1)d, where d is the
thickness of the slab. Conservation of momentum requires that the medium acquire some energy and
n
c (1 c ).
~
momentum, giving rise to a material momentum pm = By use of conservation principles,
1 ~
the photon momentum within the medium is required to be p = n c . Thus, the kinetic photon
momentum is the Abraham momentum, of which, corresponds with the Abraham momentum den-
sity. This analysis excludes other forms, such as the Livens momentum [16, 24] and the Minkowski
momentum, as being the kinetic momentum of light. In addition, other researchers have approached
this problem from differing perspective with similar results [12, 13, 25].
5
Resolution to optical momentum. In 2010, a resolution to the Abraham-Minkowski debate
was proposed [12]. With this, the proposed resolution identified equivalent relations between the ki-
netic and canonical subsystems, where the kinetic momentum is expressed as Abrahams momentum
and the canonical momentum as Minkowskis momentum. As a result, the Abraham momentum de-
scribes the overall center-of-mass translation of a material, while the Minkowski momentum renders
This thesis serves to provide insight into the mathematical modeling of relativistic electrodynamics.
Within the literature, moving media treatments have been utilized to demonstrate and analyze
electromagnetic waves within various system. Although consistent, each treatment utilizes multiple
approaches and rationales to derive, resolve, and compute analytic solutions. In the treatment
put forth by Penfield and Haus [26], for example, the relativistic methods, although exact, are
extremely complex and utilizes overly strenuous methods to describe the dynamics of a system.
In addition, other relativistic treatments employ similar mathematical manipulations in exact, but
arbitrarily complex methods, which hinder the understanding of the studied system. The results
contained herein simplify the mathematical approach, allowing for accurate modeling of complex
1.2.1 Assumptions
The theoretical calculations presented herein utilize the framework of classical electrodynamics and
the subsystem concept. The basis for each theoretical framework is developed for moving system,
and is presented in Chapter 2. The respective materials are modeled by use of the constitutive
relations, which describe the material interactions while accounting for the respective vector fields
E,
D, B,
and H.
For the purposes of this thesis, the analysis of the constitute relations are
preformed in Appendix A. In addition, the material parameters are considered to be linear, lossless,
and nondispersive.
Within this correspondence, complex notation is used to describe the monochromatic wave solu-
tions presented by each fields respective wave equation. Here, the given electric field is derived from
2 E
2 E = 0, (6)
t2
6
which is found by employing Maxwells equations. With this, 2 is the Laplacian operator, and
and are the permittivity and the permeability of the given material, respectively. Solving Eq.(6),
E = eE0 ei(krt) , (7)
Maxwells equations for the moving systems in Appendix A. Here, the used notation convention
tensors with two bars T, and scalars without any bars. Time-
expresses vectors with one bar G,
M i = 1 Re D B , where * represents the
average quantities are denoted by brackets such as hG 2
complex conjugate operator. In Chapter 5, time domain fields are related to the complex fields and
t) = Re E(z,
) .
are derived from the relation E(z,
Towards the completion of the Master of Science in Mathematics, the results in this thesis have
been or will be disseminated in referred journals and research symposiums. Below is a record of the
aforementioned works.
Journal Papers
C. J. Sheppard, B. A. Kemp, Optical pressure deduced from energy relations within relativis-
tic formulations of electrodynamics, Phys. Rev. A., Vol. 89, 013825, 2014.
C. J. Sheppard and B. A. Kemp (Faculty Advisor), Optical Pressure and Energy Relations
7
C. J. Sheppard and B. A. Kemp (Faculty Advisor), Optical Energy and Pressure from Rela-
tivistic Electrodynamics, Presented (oral) by Sheppard at the fourth annual Create @ STATE:
A Symposium of Research, Scholarship & Creativity, Jonesboro, AR (April 2014). This pre-
C. J. Sheppard and B. A. Kemp (Faculty Advisor), The kinetic of subsystem of light: A La-
grangian approach, To be presented (oral) by Sheppard at the fourth annual Create @STATE:
C. J. Sheppard and B. A. Kemp (Faculty Advisor), Balazs Thought Experiment Revisited: The
(April 2015).
8
.
2 Mathematical Framework
In order to accurately model the physical systems contained herein, it is necessary to present a
theoretical framework, of which, is used throughout this correspondence. Here, the mathematical
framework is made up of two distinct concepts, electromagnetic wave theory [18] and the subsystem
concept [26]. Electromagnetic wave theory is used to describe the physical properties of light and
how light interacts with a material substrate. The subsystem concept is used to partition a given
system into respective contribution, such as electromagnetic, mechanical, thermal, etc, where the
sum of all the energies and momenta equal zero. Using both electromagnetic wave theory and the
subsystem concept, an accurate model for moving media is developed for use in modeling moving
systems.
Here, the Minkowski, Chu, and Amperian formulations of electrodynamics are reviewed. This
serves to demonstrates how each formulations separates field and material contributions. With this,
the respective stress tensor, momentum density, energy density, and power flux for each formulation
are presented, which are used in the following sections for mathematical development and calculation.
r, t) D(
H( r, t) = J(
r, t) (8a)
t
r, t) + B(
E( r, t) = 0 (8b)
t
r, t) = (
D( r, t) (8c)
r, t)
B( = 0, (8d)
9
combine the field and material contributions within a given media [18, 26, 27]. That is, the time and
H,
space dependent field vectors E, D,
and B contain the response of the material via the constitutive
relations. These constitutive relations may take on material properties such as isotopy, bianisotropy,
losses, dispersion, etc. Here, the free charge density and free current density J are the only field
The quantities
1 1 2
fM (
r, t) = E2 H + E + J B (9a)
2 2
1
TM (
r, t) = D E + B H I DE BH (9b)
2
M (
G B
r, t) = D (9c)
define the momentum continuity equation for the Minkowski formulation. The corresponding energy
M (
r, t) = J E (10a)
SM (
r, t) = E H (10b)
1
WM (
r, t) = DE +BH , (10c)
2
A statement of charge conservation results by taking the divergence of Eq.(8a) and utilizing
Eq.(8c)
0 = J . (11)
t
Eq.(9) and Eq.(10) are derived using the predefined constitutive relations with Eq.(8). This
indicates that the desired momentum and energy terms, along with the f and terms, will depend on
the constitutive relations used in modeling the media. Thus, when using the Minkowski formulation,
we see the combination of both field and material contributions in describing the momentum and
10
2.1.2 Chu formulation
C ( EC (
r, t)
H r, t) 0 = Je (
r, t) (12a)
t
H C (r, t)
EC (
r, t) + 0 = Jh ( r, t) (12b)
t
0 EC (
r, t) = e (
r, t) (12c)
C (
0 H r, t) = h (
r, t) (12d)
within the system. The EH representation, or Chu formulation, represents the material responses
by having an equivalent electric current density Je , magnetic current density Jh , electric charge
density e , and magnetic charge density h [16]. These quantities for moving media are defined as
Je (
r, t) PC + [PC v] + JC (13a)
t
Jh (
r, t) 0 M
C + 0 [MC v ] (13b)
t
r, t) PC + C
e ( (13c)
h ( C,
r, t) 0 M (13d)
C (
where M r, t) is the magnetization, PC (
r, t) is the polarization, JC (
r, t) is the free current density,
and C (
r, t) is the free charge density of the given media. Here, the subscript C indicates that the
values involved within the Chu formulation differ from similar terms in other common formulations.
feh (
r, t)= e EC + h H C + Je o H
C Jh 0 EC
1
Teh ( C I 0 EC EC 0 H
C H C H
C
r, t) = 0 EC EC + 0 H (14a)
2
eh (
G r, t) = 0 0 EC H C , (14b)
11
and the corresponding energy continuity equations are
eh (
r, t) = Je EC + Jh H
C (15a)
Seh (
r, t) = EC H C (15b)
1 C H
C ,
Weh (
r, t) = 0 EC EC + 0 H (15c)
2
where the subscript eh denotes quantities in the Chu formulation [16, 28]. The Chu formulation
combines the bound and free electric and magnetic charges to conserve the charge within the system.
Charge conservation is written in terms of the total charge and current densities
e
0 = Je (16a)
t
h
0 = Jh . (16b)
t
r, t)
1 r, t) 0 EA ( = Jeb (
0 BA ( r, t) (17a)
t
BA (r, t)
EA (
r, t) + = 0 (17b)
t
0 EA (
r, t) = eb (r, t) (17c)
BA (
r, t) = 0 (17d)
response within a given system. The EB representation, or Amperian formulation, models the
Jeb ( A (
r, t) PA (
r, t) 0 0 M r, t) v(
r, t)
t
A (r, t) + PA ( r, t) + JA (
+ M r, t) v( r, t) (18a)
PA ( A (
r, t)
eb ( r, t) 0 0 M r, t) v(
r, t) + A (
r, t) (18b)
where PA ( A (
r, t) is the polarization, M r, t) is the magnetization, Jeb (
r, t) is the free current density,
and eb (
r, t) is the free charge density of a given medium. Here, the subscript A denotes that values
12
involved with the Amperian formulation differ from similar terms in other common formulations.
feb (
r, t) = eb EA + Jeb BA (19a)
1
Teb ( 0 EA EA + 1 1
r, t) = 0 BA BA I 0 EA EA 0 BA BA (19b)
2
eb (
G r, t) = 0 EA BA (19c)
eb (
r, t) = Jeb EA (20a)
Seb (
r, t) = 1
0 EA BA
(20b)
1
0 EA EA + 1
Web (
r, t) = 0 BA BA , (20c)
2
where subscript eb denote quantities in the Amperian formulations. Charge conservation is expressed
eb (
r, t)
0 = Jeb (
r, t) . (21)
t
The presented form of these equations makes no predictions of the exact material response in the
moving frame. Thus, the moving material can take on an description such as anisotropy, bianisotropy,
etc.
The subsystem concept expresses the energy and momentum continuity equations [26]
Wj (r, t)
j (
r, t) = Sj (
r, t) (22a)
t
j (
G r, t)
fj (
r, t) = Tj (
r, t) (22b)
t
by dividing up the total system into J subsystems, where fj is the force density, j is the power
and Wj is the energy density, for any given subsystem j. This indicates that each subsystem may
represent any of electromagnetic field, hydrostatic pressure, thermodynamic, etc. The conservation
13
of the total closed system is stated
X
j (
r, t) = 0 (23a)
j
X
fj (
r, t) = 0 (23b)
j
by closing each subsystem such that the sum of the energies and momenta for the overall system is
zero. In general, the total force or power leaving any subsystem j within a given volume is found by
integrating the force density, fj , and power density, j , respectively. To illustrate this, below the
Fe = dV Ge dA Te (24a)
t
V A
Pe = dV We dA Se , (24b)
t
V A
are found from integrating the force density and power density, respectfully. Here, the divergence
theorem was employed to reduce the electromagnetic stress tensor, Te , and power flux, Se , to a
surface integral which is integrated over surface A enclosing volume V , where V is any given volume
of the total system. Being that these results are mathematically exact, they do not depend on
the formulation used. However, these results only apply to stationary or non-moving media. To
generalize this to moving media, we must transform Eqs.(24) to accommodate moving boundaries.
This stems from how the presented partial time derivatives no longer commute with the respective
volume integrations [18]. This is solved by understanding the time rate of change of the energy and
momentum densities within a moving system. To generalize this, we define an arbitrary density
14
r, t). Taking the time derivative of the volume integration,
vector, X(
d 1 + t) dV X(t)
dV X = lim dV X(t
dt V t0 t t+t t
1 + t) + + t) dV X(t)
= lim dV X(t dV X(t
t0 t t t t
1 + t X + + t X dV X(t)
= lim dV X(t) d(vt) X(t)
t0 t t t t t t
1 X +v X
= lim dV t +v d(t) X(t) d(t) t
t0 t t t t t t
1 X + X
= lim dV t + dV ( vt) X(t) dV ( vt) t
t0 t t t t t t
1 X + X
= lim dV t + da( a vt) X(t) da(a vt) t
t0 t t t A A t
X
= dV + a v)X,
da( (25)
V t A
we find the relation presented in Eq.(25), where v is the speed of boundary interface. Rearranging
this relation and applying it to Eqs.(24), we can rewrite the electrodynamic force and power equations
as
e d
n o
Fe = a Te vG
d e
dV G (26a)
dt V
A
d
Pe = a Se vWe
d dV We . (26b)
A dt V
Here, application of Eqs.(26) render the electrodynamic subsystem for any arbitrary moving system.
Additionally, the resulting expressions are generalized, such that they describe any moving system.
Thus, by application of Eq.(25), one can transform Eqs.(26) to mathematically model a specific
At this stage, it is important to mathematically develop the time average quantities for electro-
dynamic force and power. This is accomplished by employing the time average definition
T
r, t)i = 1 r, t)
hQ( dt Q( (27)
T 0
to Eqs.(26) to find
n o
hF ie = a hTie vhGi
d e (28a)
A
e vhW ie .
hP ie = a hSi
d (28b)
A
15
Due to the generalized nature of Eqs.(26), the resulting time average values, Eqs(28), are the gen-
eralized time average expressions for any arbitrarily moving electromagnetic subsystem.
16
.
rors
In this Chapter, the optical pressures exerted on a submerged, moving perfect reflectors are mod-
eled. This serves to illustrate and resolve misunderstandings of the interpretations of the Abraham
and Minkowski momenta within the literature [15]. Here, the time average electromagnetic force
and power are modeled while having material motion, with velocity v, perpendicular to the the inci-
dent radiation. Here, the incident radiation is at normal incidence, allowing for simplification of the
system. The Chu and Minkowski formulations are used to model the Abraham and Minkowski mo-
menta for each analysis presented. Within the analysis, both PEC and PMC reflectors are utilized,
where the Fresnel reflection coefficients are given to be ei(=) = 1 and ei(=0) = 1, respect-
fully. By use of the law of conservation of energy and work-energy theorem, the system is validated
by Fe v = Pe . Additionally, the energy and momentum conservation models for the Chu and
Minkowski formulations are presented while considering reflection within a lossless, nondispersive
dielectric. Here, the subsystem method, mentioned in Chapter 2, is used to accurately demonstrate
energy and momentum conservation for moving media. Additionally, the results are compared to
The contents of this Chapter have been published in Phys. Rev. A and is cited here [1].
Here, we model the Chu and Minkowski formulations while utilizing a PEC as the perfect reflector.
The PEC has a reflection phase = so that ei = 1. The frame of reference is where the observer
sees the PEC moving with velocity v = zv, perpendicular to the surface, as seen in Fig. 1. Because
the boundary is moving, the boundary of the reflector is defined at v = zv such that z = vt.
Additionally, the mechanical work and force derived via energy and momentum conservation laws
17
, 0
i
ki
Hi Rmirror = ei
x
z = vt y
velocity
z
p
Figure 1: A plane wave propagating within a dielectric with refractive index n = /0 normally
incident onto a perfect reflector, with reflection coefficient Rmirror = ei . The American Physical
Society [1]
.
Consider an electromagnetic wave propagating within a moving dielectric half-space. The incident
Ei (
r, t) = E0 ei(ki zi t)
x (29a)
i ( E0 i(ki zi t)
H r, t) = e
yn (29b)
c0
i ( E0 n(n ) i(ki zi t)
D r, t) = x 2 e (29c)
c 0 1 n
E0 n i(ki zi t)
Bi (r, t) = y e , (29d)
c n 1
frame are
Er (
r, t) = E0 Rei(kr zr t)
x (31a)
r ( E0
H r, t) = yn Rei(kr zr t) (31b)
c0
r ( E0 n(n + ) i(kr zr t)
D r, t) = x 2 Re (31c)
c 0 1 + n
E0 n +
Br (
r, t) = y Rei(kr zr t) , (31d)
c n + 1
18
where the reflected wave vector is
n + r
kr = z , (32)
n + 1 c
and r is the reflected angular frequency. The wave vectors ki and kr are derived in Appendix A
and given in Eq.(166a) and Eq.(166b), respectfully. Additionally, the field vectors are derived in
Appendix B.
The fields inside the PEC are zero, and we ensure this by application of the boundary condition
E + v B = 0. Thus, we find
it n n+
yE0 e (1 + R) +R =0 (33)
n 1 n + 1
= kr v r = ki v i . (34)
Using Eq.(30), Eq. (32), and Eq. (103), we derive the Doppler shift formula relating r to i
1 + n
r = i . (35)
1 n
1 + n
R= . (36)
1 n
Now, we find the time average power by use of Eqs.(28b), (10b), and (10c)
1 v Re D
Re E H E + B H
hPelec i = da
A 2 4
2U0 cn2 (1 2 )
= (37)
(1 n)2
Here, U0 is defined as the energy density of the incident wave, where U0 = E02 /(2c2 0 ) for the PEC
case.
In conceptualizing the electromagnetic power flow, consider a cylinder where a unit cross section
is erected such that it is parallel to the z axis and contains a portion of the interface [23, 27]. The
defines the total electromagnetic power flow into the cylinder.
path integral of the power flux, S,
Within the cylinder, there is a decrease in stored electromagnetic energy proportional to the velocity
of the PEC. This change in electromagnetic energy stored is rendered from the velocity times the
19
path integral of the energy density. Additionally, the mechanical work is found via conservation
The radiation pressure exerted by the fields on the conductor is expressed by Eqs.(28a),(9b), and
1 v Re D
hFelec i = E + B H B
a
d Re D
A 4 2
2
(1 )
= z 2U0 n2
. (38)
(1 n)2
Here, we define the mechanical force needed to maintain the media at a constant velocity by
(1 2 )
hPmech i = hFmech i v = z2U0 n2 zv
(1 n)2
2U0 cn2 (1 2 )
= , (39)
(1 n)2
Comparing these results to similar work pertaining to the stationary media [13, 22, 28], we take
hPmech i = 0 (40a)
nE0 H0
hFmech i = z2U0 n2 = z
c
hSi i
= z2n , (40b)
c
E0 H 0
where the time average incident power is defined as hSi i = 2 .
20
3.1.2 Chu analysis
Here, we reanalyze the PEC case using the Chu formulation. To do this, we use the vector trans-
formations [26]
n o
v [E D0 ] v v (B 0 H)
EC = E + + (41a)
c2 (1
2) (1 ) 2
n o
B
v [H 0 ] v v (D 0 E)
C
H = H+ (41b)
2
c (1 ) 2 (1 ) 2
0 E)
v 0 v (B 0 H)
v
( D
PC = D 0 E +
2 2
2
(41c)
c (1 ) (1 )
C
0 M = B 0 H + v (B 0 H) v 0 v (D 0 E) , (41d)
2
c (1 )2 2
(1 )
to transform the Minkowski field vectors to the Chu field vectors for moving media. The incident
ECi (
r, t) E0 ei(ki zi t)
= x (42a)
C ( E0 n i(ki zi t)
H i r , t) =
y e (42b)
c0 1 n
E0 (1 n2 ) i(ki zi t)
PCi (
r, t) =
x 2 e (42c)
c 0 (1 n)
C (
M r, t) = y0, (42d)
i
where ki is described by Eq.(30). Additionally, the reflected Chu fields in the stationary frame are
ECr (
r, t) = E0 Rei(kr zr t)
x (43a)
C ( E0 n +
H r r , t) = y Rei(kr zr t) (43b)
c0 1 + n
E0 (1 n2 ) i(kr zr t)
PCr (
r, t) =
x 2 Re (43c)
c 0 (1 + n)
C (
M r, t) = y0 (43d)
r
21
where kr is described by Eq.(32). Using the following relation, we can validate the Chu fields by the
E = EC + 0 M
C v (44a)
H C PC v
= H (44b)
D = 0 EC + PC (44c)
B = 0 (H
C +M
C ). (44d)
it n n+
yE0 e (1 + R) +R =0 (45)
n 1 n + 1
such that R is
1 + n
R= . (46)
1 n
It is easy to see that the tangential boundary conditions, given by Eq.(33) and Eq.(45), are identical.
C = 0 to yield
This can be illustrated by applying Eq.(44a) and Eq.(44d) with magnetization M
E = EC and B = 0 H
C . As a result, the reflection coefficients are also identical, and are given in
The electromagnetic power produced by the Chu formulation is found from Eqs.(28b), (15b),
1 v
Re EC H
C
Re 0 EC EC + 0 H
C H
C
hPelec i = a
d
A 2 4
2U0 cn2 (1 2 )
= . (47)
(1 n)2
Additionally, the mechanical power needed to sustain the conductor is rendered by conservation,
To deduce the Chu radiation pressure on the conductor, we make use of Eqs.(14a), (14b), and
(28a) to find,
( )
1 v 0 0 EC H
C
hFelec i = a
d 0 EC EC 0 H
C H
C
A 2 2 t
(1 2 )
= z 2U0 n2
. (48)
(1 n)2
22
The mechanical force needed to maintain the system is defined by hFmech i = hFelec i. We validate
2U0 n2 (1 2 )
hPmech i = hFmech i v = z zv
(1 n)2
2U0 n2 c(1 2 )
= , (49)
(1 n)2
and is in agreement with the negative of Eq.(47). To compare these results to stationary media, we
take the limit as 0 to demonstrate the stationary case. In doing this, we use E0 = H0 0 to
p
E0 H 0
rewrite the energy density equation U0 = 2cn to derive
hPmech i = 0 (50a)
nE0 H0
hFmech i = z2U0 n2 = z
c
hSi i
= z2n , (50b)
c
Thus, the Chu and Minkowski formulations predict a force equal to twice the Minkowski momentum
will be imparted to the reflector at zero velocity. This is in agreement with previously published
The Minkowski and Chu formulations are modeled while utilizing a PMC as the perfect reflector.
The PMC has a reflection coefficient = 0 so that ei = 1. To derive the field vectors, we take the
[18]. Again, the frame of reference is where the observer sees the PMC moving with velocity v = zv,
perpendicular to the surface, as seen in Fig. 1. Additionally, the mechanical work and force derived
via energy and momentum conservation laws are graphed in Fig. 3 as a function of velocity.
23
25
20
Mechanical Force
15
10
0
1 0.5 0 0.5
mech i
hF
(a) The normalized mechanical force (hSi i/c)
on a PEC with the normalized
velocity ranging from 1 to 1/n.
1.5
Mechanical Work
0.5
0.5
1 0.5 0 0.5
hPmech i
(b) The normalized mechanical work hSi i
on a PEC with the normal-
ized velocity ranging from 1 to 1/n.
Figure 2: The mechanical force and work versus velocity for both the Minkowski and Chu formu-
lations are presented for the PEC. Here the normalizing factors are hSi i = E02H0 and = vc is the
normalized velocity, where n = 3.25. The Minkowski and Chu momenta are equivalent and tend to
infinity when the relation n = 1. The American Physical Society [1].
24
3.2.1 Minkowski analysis
Consider an electromagnetic wave propagating within a moving dielectric halfspace, normally inci-
dent on a PMC. The incident Minkowski fields in the stationary frame are
i (
H r, t) = yH0 ei(ki zi t) (51a)
H0 1 i(ki zi t)
Ei (
r, t) =
x e (51b)
c0 n
H0 n
Bi (
r, t) = y 2 ei(ki zi t) (51c)
c 0 n(1 n)
i ( H0 n i(ki zi t)
D r, t) = x e (51d)
c n 1
where ki is described by Eq.(30). The reflected Minkowski fields in the stationary frame are
r (
H r, t) y H0 Rei(kr zr t)
= (52a)
H0 1 i(kr zr t)
Er (
r, t) =
x Re (52b)
c0 n
H0 n +
Br (
r, t) =
y 2 Rei(kr zr t) (52c)
c 0 n(1 + n)
r ( H0 n +
D r, t) =
x Rei(kr zr t) (52d)
c n + 1
where kr is described by Eq.(32). Here, the field vectors are derived in Appendix B with respect to
the dual case of the linearly polarized electric field vector used previously.
We use the tangential boundary condition for the PMC to ensure us of the fact that there are
no fields within it. Here, the magnetic fields tend to zero on the surface of the conductor. Thus, the
v D
boundary condition H = 0 yields
n n+
xH0 eit (1 R)
R =0
n 1 n + 1
(53)
1 + n
R= . (54)
1 n
25
The electromagnetic power is found from Eqs.(28b), (10b), and (9c),
1 v Re D
H
E
+ B
H
hPelec i = a
d Re E
A 2 4
2
2U0 c(1 )
= (55)
(1 n)2
where the energy density of the incident wave is redefined by duality as U0 = H02 /(2c2 0 ). As
previous, we account for the total electromagnetic power flow at the reflector surface by the path
Additionally, the stored electromagnetic energy in front of the mirror
integral of the power flux, S.
is proportional to the speed of the mirror, and is defined by the product of the velocity of the system
with the path integral of the energy density, W . Also, by conservation, the necessary power to
sustain the system is given by Pelec = Pmech , thus allowing for the sum of the powers to be zero.
The electrical force, or radiation pressure on the reflector produced by the electromagnetic sub-
1 v Re D
hFelec i = E
+ B
H B
a
d Re D
A 2 2
2
2U0 (1 )
=
z . (56)
(1 n)2
The mechanical force required to keep the mirror moving at constant velocity is given by equating
2U0 (1 2 )
hPmech i = hFmech i v = z zv
(1 n)2
2U0 c(1 2 )
= , (57)
(1 n)2
and is in agreement with the negative of Eq.(55). To compare these results to the non-moving
hPmech i = 0 (58a)
E0 H0
hFmech i = z2U0 = zn
c
hSi i
= z2n . (58b)
c
q
Here, we use the relation H0 = E0
0 to rewrite the incident energy density as U0 = n E02cH0 and
E0 H 0
define the time average incident power for the duel case as hSi i = 2 . The Minkowski energy
26
relations give the pressure on a stationary PMC to be twice the Minkowski momentum.
Here, we reanalyze the PMC case using the Chu formulation. To do this, we utilize Eq.(41) to
transform the Minkowski field vectors to the Chu field vectors. Thus, the incident Chu fields in the
i ( n
H r, t) = yH0 ei(ki zi t) (59a)
n(1 n)
H0 1 i(ki zi t)
Ei (
r, t) = x e (59b)
c0 n
H0 (1 n2 ) i(ki zi t)
Pi (
r, t) = x
e (59c)
c n(1 n)
i (
M r, t) =
y0 (59d)
where ki is given in Eq.(30). The reflected Chu fields in the stationary frame are
r ( n+
H r, t) =
y H0 Rei(kr zr t) (60a)
n(1 + n)
H0 1 i(kr zr t)
Er (
r, t) = x Re (60b)
c0 n
H0 (1 n2 )
Pr (r, t) = x
Rei(kr zr t) (60c)
c n(1 + n)
r (
M r, t) = y0 (60d)
where kr is given in Eq.(32). We use Eq.(44) to validate the field vector quantities above. Here, we
v 0 E = 0, to
make use of the tangential boundary condition for the Chu formulation, given by H
solve the magnetic field contributions such that they are zero on the surface of the PMC. Utilizing
H0 it n n+
x e (1 + R) + R =0
n 1 n 1 + n
(61)
1 + n
R= . (62)
1 n
27
The reflection coefficients for both formulations, given by Eq.(54) and Eq.(62), are found to be the
1 v
Re EC H
C
Re 0 EC EC + 0 H
C H
hPelec i = a
d C
A 2 4
2U0 c(1 2 )
= . (63)
n2 (1 n)2
We calculate the electrical force, or radiation pressure exerted on the PMC by making use of the
1 v
hFelec i = Re 0 EC EC + 0 H
C H
C
Re 0 0 EC H
C
a
d
A 2 2
2U0 (1 2 )
=
z (64)
n2 (1 n)2
The mechanical force used in maintaining the system at a constant velocity is expressed by hFmech i =
2U0 (1 2 )
hPmech i = hFmech i v = z zv
n2 (1 n)2
2U0 c(1 2
= , (65)
n2 (1 n)2
Comparing these results to the stationary literature, we take 0 while making use of the
q
relation H0 = E0 0 to rewrite the incident energy density such that U0 = n E02cH0 to find,
hPmech i = 0 (66a)
2U0 1 E0 H0
hFmech i = z = z
n2 n c
2 hSi i
= z . (66b)
n c
Therefore, the pressure on the stationary PMC is derived from the Chu formulation to be twice the
Abraham value.
28
14 Minkowski
Chu
Mechanical Force 12
10
8
6
4
2
0
1 0.5 0 0.5
hF i
(a) The normalized mechanical force (hSmech on a PMC with the normal-
i i/c)
ized velocity ranging from 1 to 1/n.
2.5
Minkowski
Chu
2
Mechanical Work
1.5
0.5
0.5
1 0.5 0 0.5
hPmech i
(b) The normalized mechanical work (hSi i)
on a PMC with the normal-
ized velocity ranging from 1 to 1/n.
Figure 3: The mechanical force and work versus velocity for both the Minkowski and Chu formu-
lations are presented for the PMC. Here the normalizing factors are hSi i = E02H0 and = vc is the
normalized velocity, where n = 3.25. The Minkowski and Chu momenta differ corresponding to their
material contributions and tend to infinity when the relation n = 1. The American Physical
Society [1].
29
3.3 Discussion
In Sec. 3.2 and 3.3, we analyze the electromagnetic energy and momentum continuity of a plane
wave incident upon perfect reflectors submerged within a linear, lossless, and non-dispersive dielectric
medium. Both PEC and PMC reflectors are studied, which are defined as having zero tangential
electric and magnetic surface fields, respectively. In both cases, the Minkowski and Chu formulations
were applied so as to discern differences between the associated Minkwoski and Abraham momenta.
In this section, we review prior contributions and place the present work in reference to the state
of the art knowledge. The outcome of this discussion allows for logical interpretation of the two
Electromagnetic energy and momentum conservation has previously been studied for a moving
PEC in free space by Daly and Gruenberg [23]. Taking 0 in Eq.(18) of [23] for the normal
incidence case, the electromagnetic force expression of Daly and Gruenberg reduces to
1+
Felec =
z 2U0 . (67)
1
As demonstrated, the incident radiation pressure exerts a velocity dependent force onto the mirror
such that mechanical work and force are needed to close the system. Here, the closed system yields a
mechanical force Fmech = Felec necessary to keep the reflector moving at constant velocity, which
is equal and opposite to the applied electromagnetic force. Mechanical work must be put into the
system to maintain energy conservation. Additionally, we note that applying refractive index n = 1
to Eq.(38) and Eq.(48) will reduced the calculated radiation pressures to Eq.(67). Extending this
to the PMC or duel case, we find similar results for Eq.(56) and Eq.(64).
Recently, the experimental conclusion presented by the JRL experiments has been questioned
with relation to the Fresnel reflection coefficient [22]. This is due to most conventional mirrors pos-
sessing a Fresnel reflection coefficient close to -1. That is, the reflector can be appropriately described
as being a PEC with reflection phase = at normal incidence. The importance of this comes from
the fact that when considering a lossless and non-dispersive dielectric, the force calculations render
the Minkowski momentum imparted to the PEC reflector regardless of the formulation used. This
result has been verified by multiple researchers [13, 22, 28, 29]. In contrast, a PMC with reflection
phase = 0 renders either Abraham or Minkowski momenta when considering the Chu or Minkowski
formulations, respectively [13, 22, 28]. For moving media, we illustrate this by comparing Eq.(38),
Eq.(48), Eq.(56), and Eq.(64) for the given PEC and PMC models. Additionally, we summarize
the stationary case given by Eq.(40b), Eq.(50b), Eq.(58b), and Eq.(66b) in Table 2. Here, we note
30
Formulation PEC PMC
that there is no distinction between the Chu, Einstein-Laub, and Amperian formulations [16] for the
Subsequent arguments have been advanced in favor of the Minkowski momentum being observed
regardless of the type of reflector used [13]. One simple argument in favor of this conclusion considers
1 v
~ = mv 2 + ~ 1 n , (68)
2 c
where ~ is the reduced Plancks constant and m is the mass of the reflector. Solving for the reflector
momentum after reflection yields the physical solution mv = 2n~/c, which is twice the Minkowski
momentum. This conclusion is independent of the phase of the Fresnel reflection coefficient. An
explanation for this result has to do with the additional stress in the dielectric fluid resulting from the
standing wave pattern in front of the reflector [28]. Starting with the Chu formulation to represent
the kinetic subsystem of light, the additional material stresses can be derived. In the case of the
PEC reflector, the material stress at the reflector boundary is zero since the electric field is zero in
the dielectric fluid. Therefore, the total pressure at the reflector surface is purely electromagnetic
and is given by the Minkwoski momentum. However, the electric field is nonzero at the surface of
the PMC reflector, which yields a nonzero material stress at the boundary between the dielectric
fluid and the PMC. In this case, the total stress, which is the sum of the electromagnetic stress
(Abraham) and material stress (Minkowski minus Abraham) restores the Minkowski momentum
imparted to the reflector. This conclusion was questioned by Mansuripur, who based a deduction
of radiation pressure on a moving mirror from the Doppler shift [30]. The primary criticism of
the previous work by Kemp [16] was in the use of an incomplete energy balance equation. It was
claimed that the energy required to drag the dielectric liquid along with the mirror was ignored.
This issue was side stepped by suggesting that the mirror be placed outside the dielectric fluid,
which reduces to a different and trivial problem of a non-submerged mirror surrounded by vacuum
or air. In the analysis contained herein, the electromagnetic energy has been taken into account
31
within the Chu and Minkowski formulations within the framework of relativistic electrodynamics.
We have ignored, for simplicity, the additional hydrodynamic energy required to drag an object
through a fluid. However, this energy is independent of the dielectric index of refraction and the
type of reflector used. In what follows, we discuss the interpretation of the differing results from
the Chu and Minkowski formulations in terms of the electromagnetic energy required to keep the
From Table 2, it is seen that the Chu and Minkowski formulations give differing values for the
pressure on submerged reflectors at zero velocity. The basis of interpretation stems from how the
two formulations separate field and matter contributions to the energy. Consider a dielectric model
derived from the equation of motion for a bounded electron under the action of the electric field.
The Lorentz media model provides a well-established framework for establishing the relationship
between the Chu and Minkowski formulations in stationary media [18]. The Chu energy given in
Eq.(15c) gives the energy contained in the electromagnetic fields absent the energy due to the action
of the material response, which is modeled as harmonic oscillators [3133]. Addition of the resulting
material energy results in a causal material model for dielectrics [18]. In narrow frequency bands
where dispersion and losses are negligible, the energy and momentum quantities associated with the
propagating wave reduce to the Minkowski values given in Eq.(9) and Eq.(10) [3436], which have
been provided for the linear, non-dispersive dielectric considered herein. Therefore, the predictions of
momentum imparted to the PMC reflector differ between the Chu and Minkowski formulations due
to the contributions of the material polarization response to energy and momentum at the reflector
surface. This difference is absent for the PEC reflector case due to the phase of the standing wave
pattern in the dielectric in front of the reflector. That is, the electric field is zero at the surface of
the PEC and nonzero at the surface of the PMC. This analysis and conclusion is in agreement with
previous results based on a subsystem approach applied to stationary submerged reflectors [28].
Within this Chapter, both optical momentum have been derived according to the prescribed
relativistic Doppler shift for moving media. In doing this, the electromagnetic work was deduced from
energy balances within the system to find the optical pressures exerted on submerged PEC and PMC
reflectors, by which were analyzed with respect to the Minkowski and Chu formulations of Maxwells
equations. For the PEC, the pressure was found for both the Minkowski and Chu formulations
were in-fact equal, as well as consistent with the Minkowski momentum. The submerged PMC
yielded results associated with the Minkowski and Abraham momenta for the Minkowski and Chu
formulations, respectfully. These results are indicative of the separation between the optical pressure
on the mirror and dielectric medium at the interface, thus defining a clear difference in the optical
32
momenta for the dielectric and mirror while being modeled by the Chu formulation. This separation
is consistent with the causal Lorentz media model, which assigns the energy as being the Chu energy
plus the additional polarization energy resulting from the harmonic oscillators. In regards to the
optical momentum debate, the results presented herein are consistent with the quantum resolution
presented by Barnett [12] and the classical presentation by Kemp [16] in that the observable pressure
on a reflector with respect to the submerging fluid will be twice the Minkowski momentum although a
separate, and distinct, separation of momentum components based on the Chu formulation gives the
Abraham momentum plus the material response. This work contributes to the theoretical modeling
be used to calculate the force on an object immersed in a medium [37]. However, we reinforce
the position that appropriate calculations will yield equivalent physical results regardless of which
formulation of electromagnetism is employed, provided all relevant modes of momentum transfer are
included. We have illustrated this position by applying both the Chu and Minkowski formulations
to a specific physical scenario, which give rise to the Abraham and Minkowski momentum densities,
respectively. In this regard, it is recognized that the application of the Chu formulation must also
include additional energy and momentum contributions from the material response to describe the
additional momentum transfer resulting from material stresses within a dielectric. Therefore, this
approach may be considered as a theoretical basis for the study of more complex physical systems and
adds merit to the most accepted and presently growing viewpoint that accurate physical predictions
may only be obtained by considering the complete interactions between electromagnetic fields and
media.
33
.
In Chapter 3, we reviewed the optical momentum debate rendering that both Abraham and
Minkowski momenta can be modeled via electromagnetic wave theory. Although modeled by the
formulation. The implications of resolving the kinetic subsystem allows for deeper understandings
of electric, magnetic, and optical interactions with material substrates. Throughout the century,
many proposed formulation have partitioned Maxwells equations to define specific field and mate-
rial vectors quantities, allowing for arbitrarily defined kinetic subsystems. This led to the varying
The problem, however, is there can only be one accurate representation of the kinetic formulation
of electrodynamics. To resolve this conflict, energy relations are utilized to derive the kinetic elec-
tromagnetic field equations from generalized vector and scalar potentials. This is useful, being that
previous research has provided support for specific energy relations, of which corresponds to the
kin = 0 0 E
kinetic momentum [19], and thereby the kinetic momentum density G H.
By defi-
nition, this allows for the exclusion of both the Minkowski formulation and Amperian formulation,
being that the corresponding definitions for momentum density are not equal. Thus, this Chapter
demonstrates the the analysis by Lagrangian and relativistic power principles by using the kinetic
energy relations to determine the accurate dynamics of the kinetic electrodynamic subsystem.
The following sections serve as a review of the Lagrangian processes presented in continuum
mechanics [38, 39]. This develops the methods and ideologies used in the analysis for deriving the
34
4.1.1 Hamiltons variational principle
dt L. (69)
t
of a path of the possible motion of the physical system renders an extremum when evaluated along
the path of motion that is the true motion. Here, L is the Lagrangian of the system, and represents
the difference between the kinetic and potential energies of a system. So, out of the myriad of
different possibilities a system could change during time t, the true motion that does occur is
what either maximizes or minimizes Eq,(69). Thus, any variation of any system parameter (speed,
displacement, energy, etc) by some infinitesimal amount formulates the integral such that it vanishes,
dt L = 0. (70)
t
The result in Eq.(69) is called the action integral, which generates equations of motion from Hamil-
tons principle.
In consideration of continuous systems, the Lagrangian L is given from the volume integral over
L= dV L, (71)
V
where dV dX1 dX2 dX3 and L is the Lagrangian density regarding the generalized field quantities
x and
, x, x
=X
where I = 1, 2, 3 with the independent space and time variables being X 1 +X
2 +X
3
XI
and t. Using Eq.(70), a change in allows for the variation of the Lagrangian density with respect
and t, constant. Thus,
to the generalized field quantities while holding the independent variables, X
the variation of the dependent field quantities evaluated on the surface of the given volume or at
35
the beginning or ending of the time interval will vanish. Applications of this gives,
0 = L dV dt
t
V
= L dV dt
t V
L L L xj
= xj + x j + dV dt (72)
t V xj x j (xj /XK ) XK
where j and K are subscripts that acknowledge the summation over the respective variables. The
integral in Eq.(72) is a consequence of the chain rule, which allows the view or evaluation of each
ulation of the second term (by integration over time) and third term (by integration over volume)
render the variation in terms of xj , further simplifying the expression presented in Eq.(72). After
L d L d L
xj dV dt = 0 (73)
t V xj dt x j dXK (xj /XK )
The variation of xj (j = 1, 2, 3) is independent and arbitrary. Also, the time interval and system
volume are arbitrary. Thus the integral in Eq.(73) can vanish only if the coefficients of xj inside
L d L d L
0=
xj dt x j dXK (xj /XK )
d L L d L
= . (74)
dt x j xj dXK (xj /XK )
The result presented in Eq.(74) is the Lagrange equation for the Lagrangian density of a system.
As is shown in the following section, this result is used to derived the equations of motion of a given
system or subsystem. It is important to note that Lagrange equations can be applied to systems
whether or not they conserve energy or momentum, but yield relations under which energy and
The relativistic principle of virtual power is derived from the fundamental tenets of the principle of
virtual work, where the force of a system is derived via the amount of work put forth along the path of
a particle [26]. Expanding these basic tenets, the relativistic principle of virtual power utilizes prior
36
knowledge of a systems power flux, power density, and energy density giving rise to the rate of change
of pertinent physical variables. Such variables would be the electric and magnetic fields, particle
density, polarization, magnetization, etc. The expressions for these quantities must be valid even if
the material is accelerating and/or deforming. By using valid transforms (i.e Lorentz transformation)
along with prescribed definition and manipulations of the power expressions, the force density, stress
tensor, and momentum density are derived for the corresponding subsystem. Thus, the relativistic
principle of virtual power derives a systems force given that the power expressions of the system are
know, where any deviation away from the true path of the power yields a null result. Below, the
0 0 0
S0 W 0
0 0 0
v 0
v
T : (
v) + G = ( S) 2 W 0 ( v)0 + 0 (75)
t c t t
where superscript 0 denotes values within the arbitrary reference frame. Additionally, the operator
: signifies the dyadic dot product, also known as the double dot product with respect to the dyadic
matrix.
ate between the presented kinetic formulations. Within the literature, the Chu, Einstein-Laub, and
Abraham formulations possess a kinetic or field momentum density equalling the results derived by
kin , defining each formulation as a potential kinetic representation
the Balazs thought experiment, G
of light. This is due to Balazs thought experiment rendering a theoretical expression describing the
momentum density, paving the way for the modeling of the kinetic formulation. Subsequent research
0 + 0 resulting from causal energy
has found the field energy density as, Wf ield = 2 (E E 2 (H H),
and H
relations in materials [35, 36, 40]. From the presented field relations, it is assumed that E
are regarded as the respective electric and magnetic fields and represent the free space field vector
quantities. This is due to the energy and momentum density relations exclusively being expressed
and H.
in terms of the electric and magnetic field terms, E Consequently, this result limits the
study to the Chu and Einstein-Laub formulations, due to the Abraham formulation quantities being
D,
expressed in terms of E, B,
and H
vector fields. Although significant, this is unrelated being
that the Abraham formulation has been shown to be inconsistent with special relativity [4143],
which is used within the presented analysis. Here, the main concern is which formulation provides
the accurate set of physical laws prescribed by the aforementioned shared energy relations defined in
37
both the Chu and Einstein-Laub formulations, thereby providing valid physical laws for the kinetics
When considering a closed system, the Lagrangian approach produces consistent dynamical laws
from known energy relations. That is, if the energy of a system is known, one can derive the
L = LF + LI + LM (76)
where LF is the electromagnetic field Lagrangian density, LI is the field-matter interaction La-
grangian density, and LM is the matter Lagrangian density. Here, we can separate the system into
individual contributions to derive the dynamics of each individual subsystem. The electromagnetic
field Lagrangian density is defined by determining the energy and co-energy functions via dependence
on geometric variables along with Legendre transforms, leading to the expression [26]
0 2 0 2
LF = E H . (77)
2 2
By use of vector field definitions [44, 45], the electric and magnetic contributions can be rewritten
1 A
E = e C (78a)
0 t
1 C
H = m + A , (78b)
0 t
where e and m are the electric and magnetic scalar potentials, and C and A are the electric
and magnetic vector potentials, respectively. This allows for the formation of the field Lagrangian
density, leading the calculation of the kinetic subsystem without imposing any assumptions to the
scalar and vector potentials. Additionally, the field-matter interaction Lagrangian density is defined
to include all theoretically possible electric and magnetic interaction terms such that
LI = e e + Je A + m m + Jm C,
(79)
where Je and Jm are the effective electric and magnetic current densities, and e and m are the
effective electric and magnetic charge densities, respectively. This relation allows for the theoretical
modeling of magnetic charges and currents, as to not make any assumptions regarding the true
38
physical interactions of the system. Now, to allow for the accurate partitioning of field and material
subsystems, we exclude the matter Lagrangian density, LM , as it pertains only to the dynamics of the
material subsystem. That is, the matter Lagrangian describes the material subsystem independent
system, it is not necessary for the derivation of the electromagnetic fields. However, we must note
that the material subsystem must render equal and opposite force and energy expressions as derived
by the field and matter-field interaction subsystems, as stated by conservation principles. Here,
the Lagrangian density given in Eq.(76) reduces to the field and interaction Lagrangian densities,
rendering the kinetic or field Lagrangian density, denoted Lk . By use of Eqs.(77), (78), and (79),
2 2
A C
Lk (e , A, = 0
m , C) 1
e C
0
m +
1
A
2 0 t 2 0 t
e e + Je A + m m + Jm C. (80)
Thus, by employing Eqs.(74) and (80), the dynamics of the system can be derived in terms of each
scalar and vector potential. After manipulation, the collection of Lagrange equations render the
H 0 E = Je (81a)
t
+ 0 H
E
= Jm (81b)
t
0 E = e (81c)
0 H = m , (81d)
as presented in Eqs.(12). Details for the derivation of Eqs.(81) are demonstrated in Appendix C.
= 0 E
Employing field definitions D + P and B
= 0 (H
+M
) while assuming Maxwells equations,
the effective electric and magnetic current and charge densities are defined as
Je P + [P v] + J (82a)
t
Jm 0 M + 0 [M v] (82b)
t
e P + (82c)
m ,
0 M (82d)
39
equalling the results in Eqs.(13).
By taking the derived Maxwell-Chu relations, one can use standard manipulation to derive the
electromagnetic stress tensor and force density associated with the equations. However, doing this
yields an ambiguous result. This is because in modeling the kinetic force density in terms of a kinetic
H)
(E
fk = Tk 0 0 , (83)
t
only one of the three associated values with the field force density has been verified. This allows one
to add or subtract any related electromagnetic values to both the kinetic force density and stress
tensor as long as the relation hold, allowing for multiple interpretations of the electromagnetic and
material systems. This is resolved by deriving the kinetic force density by use of the relativistic
0 0
S0
0+
v W 0
Q0 = ( S) + + W 0 ( v)0 0 (84a)
c2 t t
0
v
Q0 0 0
v) + G
= T : ( 0
(84b)
t
along with prior knowledge about a given subsystem via the energy density, power flux, and power
density to derive the stress tensor and momentum density of the given subsystem within an arbitrary
inertial reference frame. Due to the division of field and material contributions within the Lagrangian
approach, the field energy density, power flux, and power density are expressed as
2 2
1 0 E + 0 H
H
k = E (85)
2 t
where the power flux and energy density are derived by standard manipulation, yielding
Sk H
= E (86a)
1 E + 0 H H
Wk = 0 E (86b)
2
k Je + H
= E Jm , (86c)
and are proven in Appendix D. With the provided energy definitions in Eqs.(86) and the relativistic
field transformations, one can employ Eq.(84) to derive the kinetic force density in terms of the kinetic
stress tensor and momentum density. To do this, the presented energy relations are transform by
40
the relations
0
E + v 0 H
= E (87a)
0
H v 0 E.
= H (87b)
Sk0 H
= E + [( E]
v 0 E) + [( H]
v 0 H) (88a)
1 1 2
v H)}]
Wk0 = 0 E 2 + 0 H 2 2 [{E (88b)
2 2 c
0k Je0 + H
= E Jm
0
. (88c)
Here, note that we only use the transformation law to first order term of the velocity. This is due
to the relative insignificant in modeling additional higher order terms, and their associated dynamic
Q0 = [ {EH + [( E]
v 0 E) + [(v 0 H) H]}]
0
0 H0 0 0 0 0 0
E
v 0 0 E 0 0 H
v EH
+ +E +H 2
c2 t t t t c2
0 0 0
0 0 0 0 Je0 + H
0 Jm
+ E E + H H ( v)0 + E 0
. (89)
2 2
Manipulation of vector and tensor identities led to the following useful relations,
E 0 0
H)
(E 0 = 0 0
E H 0 0 H E 0 J0 H 0 J0 (90a)
e m
t t
H]
0 0 H 0 )( v)0 + 0 H 0H 0 : ( v )0
[(
v 0 H) = (0 H (90b)
E]
0 0 E 0 )( v)0 + 0 E 0E 0 : ( v )0 .
[(
v 0 E) = (0 E (90c)
As can be seen, Eq.(90a) is reexpressed in terms of the electric and magnetic current densities, which
by definition contains the material polarization and magnetization vectors. Substituting Eqs.(90)
into Eq.(89) cancels the material interaction terms, thereby rendering the expression
0 0 H 0 1 0
0
v E 0 0 1 0 0 0 0 0 0
Q = 0 E E 0 H H I + 0 E E + 0 H H v )0 ,
(
t c2 2 2
(91)
where ( v)0 = ( Comparing Eq.(84b) to Eq.(91), we find the electromagnetic stress tensor
v )0 I.
41
and momentum density as
1 0 0 1 0 0
Tk = 0E
0 E E + 0 H H I 0 E 0 0 H
0H
0 (92a)
2 2
0 H
E 0
k
G = . (92b)
c2
Using the prescribed transform definitions, the Maxwell-Chu stress tensor and momentum density
are derived. Thus, the energy relations shared by both the Einstein-Laub and Chu formulations
correspond to the electrodynamics of the Chu formulation. Furthermore, the corresponding force
1
H)
(E
fk E
0 E 2 + 0 H 2 0 E 0 H
H
= 0 0
2 t
+ Je 0 H
= e E + m H Jm 0 E.
(93)
This is a significant result, being that the resulting force density corresponding to the associated
power of the electromagnetic system yields the Chu formulation, thus proving the Einstein-Laub
formulation incorrect.
Within the literature, both the Einstein-Laub and Chu formulations attempt to model the ki-
netic formulation where the partitioned material is to contain electric and magnetic dipoles. Within
each physical model, the framework of electrodynamics is employed to define the material relations,
thereby defining the electromagnetic interactions with respect to material parameters. This is well
studied for the Chu formulation, and is generalized to include dispersion, losses, and negative mate-
rial parameters via the causal Lorentz media model [35]. With this, the Chu formulation accurately
demonstrates the electromagnetic energy and momentum transfer to causal media, of which is con-
sistent with the Minkowski formulation when including material parameters. In addition, the results
from Chapter 3, demonstrate the division of field and material contributions, validating the separa-
tion of field and material contributions via conservation principles. However, the approach used in
Chapter 3 is also valid for the Einstein-Laub formulation. This stems from both formulations shar-
ing energy relations, of which, were used to verify each force expression. The main difference in the
two formulations arise from the modeled force density expressions, which stem from the differences
42
between both (Chu and Einstein-Laub) stress tensor expressions,
fChu
1 E
0 E 2 + 0 H 2 0 E 0 H
H 0 0 (E H)
= (94a)
2 t
fEL
1
0 E 2 + 0 H 2 DEB 0 0 (E H) .
H
= (94b)
2 t
Here, fChu and fEL represent the Chu and Einstein-Laub force densities, respectively. Within the
literature, the Einstein-Laub formulation has been shown to have enate discrepancies when modeling
dipolar forces, which was first pointed out in the early to mid 1900s [46]. Over the course of the
study, Einstein replied to a inquiring researcher stating, It has long been known that the values I
had derived with Laub at the time are wrong; Abraham, in particular, was the one who presented
this in a thorough paper. The correct strain tensor has incidentally already been pointed out by
Minkowski [47]. Although mathematically correct, the Minkowski stress tensor corresponds to the
Minkowski formulation, which represents the canonical subsystem and not the kinetic subsystem.
However, one may partition the Maxwell-Minkowski equations to derive electromagnetic formulations
that differentiate between field and material contributions and are consistent with electromagnetic
wave theory. However, multiple perspectives have led a more fundamental debate as to the relevance
When reviewing the literature, three viewpoints are taken when studying the kinetic formulation
2. A kinetic subsystem cannot be uniquely identified within a material due to the inability to
As can be seen, viewpoints 1 and 2 yield trivial results, which stem from the lack of uniqueness of
formulation or lack of ability for verification. That is, either the kinetic subsystem doesnt exist
or the kinetic subsystem is defined by values which satisfy Maxwells equations, thereby defining a
particular formulation of electrodynamics. However, using viewpoint 1 can lead one to ambiguous
relations without any validation, and can lead to multiple postulated formulations without any
justification. Conversely, viewpoint 3 determines the kinetic formulation based on uniqueness within
theoretical and experimental truths, resolving the kinetic formulation to be consistent with Maxwells
equations. Due to the trivial nature of viewpoints 1 and 2, the presented research utilizes viewpoint
43
3 along with valid physical principles and known mathematical methods to accurately partition
both field and material subsystems allowing for the derivation of the true kinetic formulation of
light. Thus, the analysis presented describes the kinetic field dynamics, producing the force and
power expressions for the electromagnetic subsystem in terms of energy and momentum density
relations [19, 35, 36]. This allows for the derivation of Maxwells equations in terms of the shared
energy relations, as provided by both formulations. In doing this, all assumptions where stated and
are consistent with Maxwells equations. By use of the relativistic principle of virtual power, we
formulate the corresponding electromagnetic force density responsible for the dynamics tied to the
shared energy relations. The partitioning of field and material responses allows for the derivation of
the kinetic subsystem, which produced values in terms of the Chu formulation, thereby rendering
44
.
Material
In this Chapter, we revisit the Balazs box thought experiment while utilizing the framework of
relativistic electrodynamics. This serves to expand on the prescribed ideas of the kinetic formulation
produced in Chapter 4. Within this approach, we utilize the force and power expressions of normally
incident light on a magneto-dielectric slab moving with velocity v in the positive z direction. The
Minkowski, Chu, and Amperian formulations are used within the framework to model the electrody-
namics of the system in terms of each formulation. Due to the sophistication of the thought problem,
analysis is preformed for both the time average and time varying cases. This allows for a complete
understanding of the system and how the electromagnetic and material contributions change with
respect to both space and time. Thus, this Chapter renders significant results for understanding the
differences between the formulations of electrodynamics and their associated material contributions.
Here, we consider the problem of an electromagnetic wave normally incident from vacuum onto
a rigid, lossless, non-dispersive, isotropic magneto-dielectric slab, where an observer sees the slab of
material moving along the z axis within vacuum, as in Fig 4. The slab has a given thickness, d,
with the boundaries defined at z = vt and z = vt + d. The wave solutions are derived form the
positive direction. Within this section, we analyze the force and power for the time average for the
Here, we utilize the Minkowski formulation to first present the time average analysis for the
moving slab of material. Thus, consider an electromagnetic wave normally incident on a moving
45
q
Figure 4: A plane wave normally incident on a magneto-dielectric with refractive index n = 0 0 ,
moving with velocity v = zt.
Ei E0 ei(ki zi t)
= x (95a)
E0 i(ki zi t)
Hi = y e (95b)
c0
ki = zki (95c)
i
ki = , (95d)
c
where subscript i represents the incident field relations. Here, we note that the presented vacuum
field relations are derived by allowing n 1, 0 , and 0 , in the derived field relations
Er = E0 Rei(kr z+r t)
x (96a)
r E0
H =
y Rei(kr z+r t) (96b)
c0
kr =
z kr (96c)
r
kr = , (96d)
c
where R represents the reflection coefficient and subscript r denotes the reflected field relations, of
which are derived in vacuum. Within the moving slab, there are two wave coefficients denoted A
and B. With this, coefficient A represents the amplitude of the wave propagating in the positive
z axis where coefficient B represent the wave propagating in the negative z axis. The Minkowski
46
fields for the positive propagating wave within the material are
Ea = AE0 ei(ka za t)
x (97a)
E0 n +
Ba = y Aei(ka za t) (97b)
c 1 + n
a E0 n(n + ) i(ka za t)
D = x 2 0 Ae (97c)
c 1 + n
a n
H = y 0 AE0 ei(ka za t) (97d)
c
ka = zka (97e)
a n+
ka = na , na = . (97f)
c 1 + n
Additionally, the Minkowski fields for the negative propagating wave within the material are
Eb E0 Bei(kb z+b t)
= x (98a)
E0 n
Bb =
y Bei(kb z+b t) (98b)
c 1 n
b E0 n(n ) i(kb z+b t)
D = x
2 0 Be (98c)
c 1 n
b n
H = y 0 BE0 ei(kb z+b t) (98d)
c
kb =
z kb (98e)
b n
kb = nb , nb = . (98f)
c 1 n
The resulting relations for electromagnetic field within moving media are derived in Appendix B.
Also, the wave vectors are derived in Appendix A. The transmitted Minkowski fields are
Et T E0 ei(kt zt t)
= x (99a)
t E0
H = y T ei(kt zt t) (99b)
c0
kt = zkt (99c)
t
kt = (99d)
c
where T represents the transmission coefficient and subscript t denotes the transmitted wave rela-
tions.
Here, we use the standard boundary conditions to evaluate the field relations at each boundary
47
for the tangential components of the electromagnetic wave. The boundary conditions are given as,
z (E + v B)
= 0, (100a)
v D)
z (H = 0. (100b)
Applying Eqs.(100) at both z = vt and z = vt + d, we find the system of equations used for solving
1 2 1 2
(1 ) + R(1 + ) = A +B (101a)
1 + n 1 n
1 2 1 2
An Bn
(1 ) R(1 + ) = 0 (101b)
0r 1 + n r 1 n
1 2 1 2
A eikA d + B eikB d = T (1 )eikt d (101c)
1 + n 1 n
1 2 1 2
An Bn
eikA d eikB d = T (1 )eikt d . (101d)
0r 1 + n 0r 1 n
Using algebraic techniques, the solutions for coefficients A, B, R, and T are found to be,
Viewing Eqs.(102), we can see that the Doppler shift is present in all but the transmission
coefficient, T . That is, every coefficient expresses a term (1 x), where x = 1, n. To understand
this relation, and how the Doppler shift effects the calculations, we take advantage of phase matching
the wave at each boundary. Here, the concept of phase matching states that all waves interacting at
a specific boundary have identical phases at the given boundary. This implies the following phase
48
relations,
ki v i = kr v r = ka v a = kb v b = kt v t . (103)
Manipulation of Eq.(103) allows us to find the incident, reflected and/or transmitted angular fre-
quencies for each region in terms of the incident angular frequency, i . Thus,
1
r = i (104a)
1+
1 + n
a = i (104b)
1+
1 n
b = i (104c)
1+
t = i (104d)
Although, the systems energy transfer comes from the sum of waves in each region. This implies
that when the system is moving, there is a change in the standing wave frequency on one side of the
slab, which is proportional to the velocity of the slab, and thereby yields a net change in energy and
momentum within the incident and transmitted regions around the slab of material. This difference
in energy and momentum indicates a force and power difference, of which, produces a non-zero net
By use the force and power expressions derived in Eqs.(28a) and (28b), the time average electro-
dynamic relations are calculated for the closed system. Using the prescribed Minkowski expressions
derived in Chapter 2, the time average stress tensor, momentum density, energy density and power
1 v Re D
hFelec i = E + B H B
a
d Re D
A 4 2
2 h
E i
z 2 0 1 + |R|2 |T |2 1 |R|2 |T |2
=
c 0
2
E02 (1 ) 2 n2 02 r (1 cos())
= z 2 . (105)
2c 0 (1 + ) (n4 + 6n2 2 + 04 ) (n2 02 )2 cos()
r r r
This result is significant, being that there are no material contributions from the slab within the time
average force expression. That is, when calculating the force from the slab, calculations from both
49
Region hTzz i |hGi|
E02 2
E02 2
0 2c2 0 1 + |R| 2c3 0 1 |R|
E02 n 2 2
E02 n 2 2 2 2
1 2c2 0 0r na |A| + nb |B| 2c3 0 0r na |A| nb |B|
E02 2 E02 2
2 2c2 0 |T | 2c2 0 |T |
Region hW i |hSi|
E02 2
E02
0 2c2 0 1 + |R| 2c0 1 |R|2
E02 n 2 2
E02 n 2
1 2c2 0 0r na |A| + nb |B| 2c2 0 0r |A| |B|2
E02 2 E02 2
2 2c3 0 |T | 2c0 |T |
Table 3: The derived values for the Minkowski subsystem. The Minkowski stress tensor and mo-
mentum density are expressed in (a) and the Minkowski power flux and energy density are expressed
in (b), for each region of interest.
boundaries of the slab cancel rendering the time average electromagnetic force in vacuum quantities.
Mathematically, this is due to invoking phase matching conditions such that when the applying the
definitions for the electromagnetic force, the complex oscillation terms produced by the fields cancel.
1 v
Re EC H
C
Re 0 EC EC + 0 H
C H
hPelec i = a
d C
A 2 4
E02 h i
1 |R|2 |T |2 1 + |R|2 |T |2
=
c0
2
E02 (1 ) 2 n2 02
r (1 cos())
= (106)
2c0 (1 + ) (n4 + 6n2 2 + 04 ) (n2 02 )2 cos()
r r r
and is void of material contributions. Using the work-energy theorem hF i v = hP i, we check the
2
E02 (1 ) 2 n2 02
r (1 cos())
hFelec i v = z zc
2c2 0 (1 + ) (n4 + 6n2 2 + 04 ) (n2 02 )2 cos()
r r r
2
E02 (1 ) 2 n 2
02
(1 cos())
r
=
2c0 (1 + ) (n4 + 6n2 2 + 04 ) (n2 02 )2 cos()
r r r
= hPelec i. (107)
It is easily observed that substituting Eq.(107) into the conservation relation, the electromagnetic
50
force and power are equal and consistent within the system,
hFelec i v hP i = 0
hP i hPelec i = 0. (108)
Additionally, it is easy to see that the two subsystems, electromagnetic and mechanical, are equal
and opposite such that the mechanical force and power are needed to sustain system constraints
(such as special relativity). This result invokes the given relations, hFelec i = hFmech i and hPelec i =
hPmech i to satisfy global energy and momentum conservation. Thus, the moving time average
Although we used the Minkowski formulation, we could have easily used any of the other leading
formulations (i.e Chu, Amperian, etc...) for this calculation. The reason for this stems from how
the time average force values are calculated. That is, the non-zero electromagnetic force and power
values are rendered in the vacuum regions only, yielding the time average electromagnetic force
inside the material as a null value. This implies that the time average force calculations above are
the result of the electromagnetic stress and change in momentum density around the material as it
traverses free space. A plot of the force and power relation are demonstrated in Fig 5.
In the previous section, we studied the time average forces which yielded a result that corresponds
to all formulations. However, the main interest is to study the differing formulations and how
each formulation models the electromagnetic and material subsystems as light propagates within
the moving medium. Due to the significant, yet trivial nature of the results provided by the time
average analysis, expansion to the time varying frame allows for a more complete analysis of how
the field and material interactions occur, thereby demonstrating the physical and mathematical
differences between each formulation. Thus, this section expands the previous analysis so to study
the electromagnetic forces in and round the moving slab with respect to both space and time. In
doing so, we take an identical perspective as seen in Fig 4. This allows for the application of
each formulation of interest, and to analyze how each formulation separates the field and material
To expand the arguments into the time varying frame, we make use of the definition [18],
t) = Re E(z,
)
E(z, (109)
51
3.5e11
3e11
2.5e11
2e11
1.5e11
1e11
5e12
0e00
0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1
(a) The electromagnetic force on a moving slab, with the normalized ve-
locity ranging from 0.8 to 1. Here, hF ianalytic represents the analytic
expression of the force, where hF icomp represents the solution including
coefficients R and T .
1e03
0e00
1e03
2e03
3e03
4e03
5e03
6e03
7e03
8e03
0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1
(b) The electromagnetic power on a moving slab, with the normalized ve-
locity ranging from 0.8 to 1.
Figure 5: The electromagnetic force and work versus velocity for all formulations are presented
for a moving magneto-dielectric slab of thickness d = 0 /4n. Here 0 = 640nm, r = 5, r = 3,
n = r r , with = vc as the normalized velocity.
52
) represents values from the time harmonic electromagnetic field relations used in the
where E(z,
Sec 5.1. This defines the electromagnetic fields in the time domain, where the fields vary with respect
to both space and time. In viewing the time average field expressions, it is found that the complex
exponential and the respective wave coefficients are the only terms to change. Thus, we expand each
complex oscillation and wave coefficient terms for each respective wave. As a result,
I = cos(ki z i t) (110)
are the associated oscillation terms for each respective field solution. Here, subscript R and I denote
the real and imaginary values of the respective field coefficients. Thus, the real and imaginary wave
( 1) n4 4r (cos() 1)
RR = 2 (115a)
( + 1) (n2 2r ) cos() ( + 1) (n4 + 6n2 2r + 4r )
2n( 1)r (n r )(n + r ) sin()
RI = (115b)
2
( + 1) n4 (n2 2r ) cos() + 6n2 2r + 4r
2r (n + 1)(n + r ) (n + r )2 (n r )2 cos()
AR = (115c)
2
( + 1) 2 (n2 2r ) cos() + (n r )4 + (n + r )4
2r (n + 1)(n r )2 (n + r ) sin()
AI = (115d)
2
( + 1) 2 (n2 2r ) cos() + (n r )4 + (n + r )4
2r (1 n)(n r ) (n + r )2 cos() (n r )2
BR = (115e)
2
( + 1) 2 (n2 2r ) cos() + (n r )4 + (n + r )4
2r (1 n)(n r )(n + r )2 sin()
BI = (115f)
2
( + 1) 2 (n2 2r ) cos() + (n r )4 + (n + r )4
4nr (n + r )2 cos(A ) (n r )2 cos( A )
TR = 2 (115g)
2 (n2 2r ) cos() + (n r )4 + (n + r )4
4nr (n r )2 sin( A ) + (n + r )2 sin(A )
TI = 2 (115h)
2 (n2 2r ) cos() + (n r )4 + (n + r )4
53
5.2.1 Continuity equations for the moving slab
To accurately model the given system, it is advantageous to modify Eqs.(26a-26b) from the gener-
alized force and power expressions. This is done by applying Eq.(25) to derive the necessary relations
for each region of interest. Using Fig. 4, the appropriate continuity equations are given as
o e
n
G
Fe
= a Te vGe +
d a vGe
d dV (116a)
t
A A V
We
a Se vWe +
Pe = d a {
d v We } dV .. (116b)
A A V t
To illustrate this, we divide Eq.(116a) into contributions within each region and at each boundary.
This renders
n
o d z1
FR0 =
a T vG
d dz G (117a)
R0 dt z0
+
n
o d z1
FB1 =
a T vG
d dz G (117b)
B1 dt z1
n
o d z2
FR1 =
a T vG
d dz G (117c)
R1 dt z1+
+
n
o d z1
FB1 =
a T vG
d dz G (117d)
B1 dt z1
z3
d
n o
FR1 = a T vG
d
dz G. (117e)
R2 dt z2+
where
Here, superscripts + and denote the evaluation of the given expression on the positive or negative
z1+ z1+
d = d = 0.
dz G dz G (119)
dt z1 dt z1
54
This is due to a lack in volume to integrate with respect to. Thus, the by application of Eq.(25)
into Eq.(117), the resulting force expression for each respective region is
z1
G
FR0 = a T
d dz (120a)
R0 z0 t
n o
FB1 = a T vG
d (120b)
B1
z2
G
FR1 = a T
d dz (120c)
R1 z1+ t
n o
FB2 = a T vG
d (120d)
B2
z3
G
FR2 = a T
d dz . (120e)
R2 z2+ t
o in
n G
Fe a Teout vG
out e
+ FR0 + FR2 ,
in
= d e + d
a v
G e dV (121)
A A V t
where superscript out represents field relations outside the slab of material while superscript in
represents field relations inside the slab of material. Here, identical manipulations occur when
solving the the total power expression. Thus, the total power expression is given as
Wein
a Seout vWeout + a vWein
Pe = d d dV + PR0 + PR2 . (122)
A A V t
The expressions for the force and power in regions 0 and 2 are discussed in the latter sections of
this Chapter.
Here, we present the time domain Minkowski fields derived from Eq.(109), along with the time
harmonic fields in Eq.(95-99). Thus, the time domain incident Minkowski fields are
Ei = E0 I
x (123a)
i E0
H = y I (123b)
c0
ki = zki (123c)
i
ki = . (123d)
c
55
The reflected time domain Minkowski fields are
Er E0 R
= x (124a)
r E0
H =
y R (124b)
c0
kr =
z kr (124c)
r
kr = . (124d)
c
The time domain Minkowksi fields for the positive propagating electromagnetic wave within the
material are
Ea = E0 A
x (125a)
E0 n +
Ba = y A (125b)
c 1 + n
a E0 n(n + )
D = x 2 0 A (125c)
c 1 + n
a n
H = y 0 E0 A (125d)
c
ka = zka (125e)
a n+
ka = na , na = . (125f)
c 1 + n
The time domain Minkowski fields for the negative propagating electromagnetic wave within the
material are
Eb E0 B
= x (126a)
E0 n
Bb =
y B (126b)
c 1 n
b E0 n(n )
D = x
2 0 B (126c)
c 1 n
b n
H = y 0 E0 B (126d)
c
kb =
z kb (126e)
b n
kb = nb , nb = (126f)
c 1 n
56
Region Tzz |G|
E02 2 2
E02 2 2
0 c2 0 I + R c3 0 I R
E02 n 2 2
E02 n 2 2 2 2
1 c2 0 0r na A + nb B c3 0 0r na A nb B
E02 2 E02 2
2 c2 0 T c2 0 T
Region W |S|
E02 2 2
E02 2 2
0 c2 0 I + R c0 I R
E02 n 2 2
E02 n 2 2
1 c2 0 0r na A + nb B c2 0 0r A + B
E02 2 E02 2
2 c3 0 T c0 T
Table 4: The derived values for the time domain Minkowski subsystem. The Minkowski stress tensor
and momentum density are expressed in (a) and the Minkowski power flux and energy density are
expressed in (b), for each region of interest.
Et E0 T
= x (127a)
t E0
H = y T (127b)
c0
kt = zkt (127c)
t
kt = . (127d)
c
Here, we utilize the Minkowski stress tensor, momentum density, energy density, and power
flux as previously defined to derive the values for the time varying force and power of the electro-
magnetic subsystem. Eqs.(121) and (122) are used to derive the force and power provided by the
electromagnetic subsystem. Additionally, Table 4 demonstrates the values utilized for calculation.
E02
FM = z [(I 2 + R2 T 2 ) {I 2 R2 T 2 }]
c2 0
E02 n
v n2a A2 (z1+ ) n2b B 2 (z1+ ) v n2a A2 (z2 ) n2b B 2 (z2 )
+
z 3 0
c 0 r
A2 B 2
2 2
na dz nb dz , (128)
R1 t R1 t
where subscript M denotes the Minkowski electrodynamic subsystem. Here, FR0 and FR0 render a
null result. This is due to both stress tensor and momentum density integration being equal and
57
1e03
8e04
6e04
4e04
2e04
0e00
2e04
4e04
6e04
8e04
0e00 1e15 2e15 3e15 4e15 5e15 6e15 7e15 8e15
opposite within each region. That is, integration of the values rendered by the stress tensor are
equal to the negative of the values rendered by the momentum density, thereby canceling to provide
a zero force. This result also follows physical intuition, being that there is no electromagnetic force
exerted on a vacuum. However, there is a nonzero electromagnetic force exerted within and on the
E02
PM = [(I 2 R2 T 2 ) + {I 2 + R2 T 2 }]
c0
E02 n
v na A2 (z1+ ) + nb B 2 (z1+ ) v na A2 (z2 ) + nb B 2 (z2 )
+ 2 0
c 0 r
A2 B 2
na dz + nb dz . (129)
R1 t R1 t
Similarly, PR0 and PR2 both render null results due to cancelations within the terms of both expres-
sions. Again this is result agrees with intuition, due to the fact the electromagnetic power does no
work on vacuum.
Simplification of Eqs.(128) and (129) is somewhat impractical, and is due to the complex nature
of the real and imaginary wave coefficients, constituting the wave relations. Alternatively, software is
is shown in Fig 6 for validation. As can be seen, the validation is proved. Thus, the Minkowski
formulation satisfies the electrodynamic subsystem. In addition, we note that to satisfy the overall
58
system, there is an equal and opposite mechanical force and power such that FM = Fmech and
PM = Pmech . This allows for global conservation and satisfies special relativity.
The Chu fields are derived by using Eq.(41) and the Minkowski time domain fields. In addition,
the Chu boundary conditions render identical results to that of the Minkowski formulation, thereby
allowing for the use of Eqs.(110) and (115) in reexpressing the Chu field relations. Thus, the time
ECi E0 I
= x (130a)
Ci E0
H = y I (130b)
c0
ki = zki (130c)
i
ki = . (130d)
c
ECr = E0 R
x (131a)
Cr E0
H =
y R (131b)
c0
kr =
z kr (131c)
r
kr = . (131d)
c
The time domain Chu fields propagating in the positive direction within the material are
0r + n
ECa = x
E0 A (132a)
0r (1 + n)
Ca n + 0r E0
H = y 0 A (132b)
r (1 + n) c0
n2 0r E0
PCa = x 0 A (132c)
r (1 + n) c2 0
Ca n(0 1) E0
0 M = y 0 r A (132d)
r (1 + n) c
ka = zka (132e)
a n+
ka = na , na = (132f)
c 1 + n
59
0r +n n+0r
where cea 0r (1+n) and cha 0r (1+n) . The time domain Chu fields propagating in the negative
0r n
ECb = x
0
E0 B (133a)
r (1 n)
Cb n 0r E0
H = y 0 B (133b)
r (1 n) c0
n2 0r E0
PCb = x 0 B (133c)
r (1 n) c2 0
Cb n(0 1) E0
0 M = y 0 r B (133d)
r (1 n) c
kb =
z kb (133e)
b n
kb = nb , nb = (133f)
c 1 n
0r n n0r
where ceb 0r (1n) and chb 0r (1n) . The transmitted time domain Chu fields are
ECt = E0 T
x (134a)
Ct E0
H = y T (134b)
c0
kt =
z kt (134c)
t
kt = . (134d)
c
Here, the Chu stress tensor, momentum density, energy density, and power flux are utilized
to derive the the values for the time varying force and power for the electromagnetic subsystem.
Eqs.(121) and (122) are used to derive the electromagnetic force and power in terms provided by
the Chu formulation. For simplicity in calculation, Table 5 demonstrates the derived values for
calculating the desired time varying electromagnetic force and power contributions.
E02
FC = z [(I 2 + R2 T 2 ) {I 2 R2 T 2 }]
c2 0
E2 n
z 3 0 0 v (cea A(z1+ ) + ceb B(z1+ ))(cha A(z1+ ) chb B(z1+ ))
+
c 0 r
v(cea A(z2+ ) + ceb B(z2+ ))(cha A(z2+ ) chb B(z2+ ))
A2 B 2
AB
cea ceb dz + (cea chb ceb cha ) dz + cha chb dz . (135)
R1 t R1 t R1 t
where subscript C denotes the Chu electromagnetic subsystem. Here, we note that FR0 = FR2 = 0
as previously specified, rendering Eq.(135) as the electromagnetic force within and at the boundaries
60
Region Tzz |G|
E02 2 2
E02
0 c2 0 I +R c3 0 I 2 R2
E02 n E02 n
1 c2 0 0r (cae A + cah B)2 + (cbe A cbh B)2 c3 0 0r [(cae A + cah B)(cbe A cbh B)]
E02 2 E02 2
2 c2 0 T c2 0 T
Region |W | |S|
E02 2 2
E02
0 c2 0 I +R c0 I 2 R2
E02 n E02 n
1 c2 0 0r (cae A + cah B)2 + (cbe A cbh B)2 c0 0r [(cae A + cah B)(cbe A cbh B)]
E02 2 E02 2
2 c3 0 T c0 T
Table 5: The derived values for the time domain Chu subsystem. The Chu stress tensor and
momentum density are expressed in (a) and the Chu power flux and energy density are expressed
in (b), for each region of interest.
of the given medium. Additionally, the time varying Chu electromagnetic power is given as
E02
PC = [(I 2 R2 T 2 ) + {I 2 + R2 T 2 }]
c0
E02 n
2 2
+ 2 v cea A(z1+ ) + ceb B(z1+ ) v cha A(z2 ) chb B(z2 )
c 0 0r
A2 B 2
2 2 AB 2 2
(cea + cha ) dz 2(cea ceb cha chb ) dz (ceb + chb ) dz .
R1 t R1 t R1 t
(136)
where PR0 = PR1 = 0 as shown. Here, due to the sophistication of the terms involved, it is more
practical and equally valid to plot the conservation relations F v = P , and is done in Fig 7. As
is observed, the Chu formulation doesnt satisfy the conservation theorem for the electromagnetic
subsystem. This is due to the respective electromagnetic fields interacting with the optical medium
causing a portion of the electromagnetic energy to transfer to the material. This induces the Chu
dipolar model to oscillate with respect to the presented field relations, producing both kinetic and
potential energies within the material [35]. Consequently, this causes the electromagnetic force to
deform the material, allowing for the conservation of energy and momentum to break down. Thus,
the electromagnetic energy transferred to the oscillating dipoles causes the observed differences
61
5e04
4e04
3e04
2e04
1e04
0e00
1e04
2e04
3e04
4e04
5e04
0e00 1e15 2e15 3e15 4e15 5e15 6e15 7e15 8e15
The Amperian time domain fields are derived from the Chu time domain fields by employing the
relations [26],
EA = EC + 0 M
v (137a)
BA = C + 0 M
0 H (137b)
PA = PC (137c)
A
M = PC . (137d)
Additionally, the Amperian boundary conditions render identical results to that of the Minkowski
formulation, thus allowing for the use of Eqs.(110) and (115) in defining the Amperian field relations.
EAi = E0 I
x (138a)
Ai E0
H = y I (138b)
c0
ki = zki (138c)
i
ki = , (138d)
c
62
The time domain reflected Amperian fields are
EAr E0 R
= x (139a)
E0
BAr =
y R (139b)
c0
kr =
z kr (139c)
r
kr = . (139d)
c
The time domain Amperian fields propagating in the positive direction within the material are
EAa E0 A
= x (140a)
n + E0
BAa = y A (140b)
1 + n c0
n2 0r E0
PAa = x 0 A (140c)
r (1 + n) c2 0
Aa n(0 1) E0
0 M = y 0 r A (140d)
r (1 + n) c
ka = zka (140e)
a n+
ka = na , na = (140f)
c 1 + n
The time domain Amprian fields propagating in the negative direction within the material are
EAb E0 B
= x (141a)
Ab n E0
H =
y B (141b)
1 n c0
n2 0r E0
PAb = x 0 B (141c)
r (1 n) c2 0
Ab n(0 1) E0
0 M = y 0 r B (141d)
r (1 n) c
kb =
z kb (141e)
b n+
kb = nb , nb = (141f)
c 1 + n
63
Region Tzz |G|
E02 2 2
E02
0 c2 0 I +R c3 0 I 2 R2
E02 n E02 n
1 c2 0 0r A + na B)2 + (A nb B)2 c3 0 0r [(A + na B)(A nB B)]
E02 2 E02 2
2 c2 0 T c2 0 T
Region |W | |S|
E02 2 2
E02
0 c2 0 I +R c0 I 2 R2
E02 n E02 n
1 c2 0 0r (A + na B)2 + (A nb B)2 c0 0r [(A + na B)(A nb B)]
E02 2 E02 2
2 c3 0 T c0 T
Table 6: The derived values for the time domain Amperian subsystem. The Amperian stress tensor
and momentum density are expressed in (a) and the Amperian power flux and energy density are
expressed in (b), for each region of interest.
EAt E0 T
= x (142a)
At E0
H = y T (142b)
c0
kt =
z kt (142c)
t
kt = . (142d)
c
Here, the Amperian stress tensor, momentum density, energy density, and power flux are used
to derive the values for the time varying force and power for the electromagnetic subsystem. Thus,
Eqs.(121) and (122) are used to derive the force and power in terms provided by the electromagnetic
E02
FA = z [(I 2 + R2 T 2 ) {I 2 R2 T 2 }]
c2 0
E02 n
+
z 3 v (A(z1+ ) + B(z1+ ))(na A(z1+ ) nb B(z1+ ))
c 0 0r
v(A(z2+ ) + B(z2+ ))(na A(z2+ ) nb B(z2+ ))
A2 B 2
AB
na dz + (nb na ) dz + nb dz . (143)
R1 t R1 t R1 t
Here, we note that FR0 = FR2 = 0 as previously specified, allowing for the presented result. Addi-
64
4e04
3e04
2e04
1e04
0e00
1e04
2e04
3e04
0e00 1e15 2e15 3e15 4e15 5e15 6e15 7e15 8e15
E02
PA = [(I 2 R2 T 2 ) + {I 2 + R2 T 2 }]
c0
E02 n
2 2
+ 2 v A(z1+ ) + B(z1+ ) v na A(z2 ) nb B(z2 )
c 0 0r
A2 B 2
2 2 AB 2
(1 + na ) dz 2(1 na nb ) dz (1 + nb ) dz .
R1 t R1 t R1 t
(144)
where PR0 = PR1 = 0 as previously shown. Here, due to the sophistication of the mathematical
subsystem, and is demonstrated in Fig 8. As is observed, the Amperian formulation doesnt satisfy
the electromagnetic subsystem. This is due to the electromagnetic force deforming the rigid body of
the medium, allowing for the conservation of energy and momentum to break down [35]. Similar to
the Chu formulation, the Amperian material subsystem aligns itself with the electric and magnetic
field relations, imposing an oscillating dipolar motion. As a result, the oscillating motion produces
both kinetic and potential material energies within the material. This indicates that a portion of
the electromagnetic energy is transferred to the material, which in turn causes a net difference in
the electromagnetic energy contained within the pure electromagnetic field, thereby causing the
65
5.3 Discussion
The Balazs box thought experiment derived the kinetic momentum density, paving the way for
the formulation of the kinetic subsystem. Over the years, many researchers have revisited this
thought experiment resulting in the Abraham momentum and Abraham momentum density. Within
this Chapter, we utilize the presented ideas of the box thought experiment while employing the
framework of relativistic electrodynamics for both the time average and time varying cases. In
considering the time average case, the electromagnetic force and power expressions were shown to
be valid and independent of the formulation used. That is, when applying any given electromagnetic
formulation, the force and power results were correct, yet ambiguous. As is shown in Eq.(107), both
electromagnetic force and power expressions are rendered in vacuum quantities, leading to ambiguity
between each formulation attempting to model the system electromagnetically. In viewing the total
force density,
n o
hftotal i = hTelec i + hTmat i + hTmech i + mat i + hG
mech i = 0,
hGelec i + hG (145)
t
it is seen that there are three main contributions to the overall system, where hTelec i and hG
elec i
are the electromagnetic stress tensor and momentum density, hTmat i and hG
mat i are the material
and momentum density, respectively. As one would expect, modeling the system while utilizing any
electrodynamic formulation allows for different values to be partitioned such that different contribu-
tions of the overall system are divided into the field and material subsystems. Although significant,
the field expressions within the material cancel, leaving no mathematical ability to differentiate the
system in terms of a specific formulation. In addition, Eq.(23b) mathematically states that the sum
of the force densities must be zero, and has been shown to only include the respective electromag-
netic and mechanical force and power contributions. This implies that the time average material
subsystem,
n o
hfmat i = hTmat i +
hGmat i = 0 (146)
t
gives a zero net force, where fmat is defined as the material force density. Physically, this states that
however one idealizes the material subsystem, the time average material contribution adds no value
to the system, and doesnt deform or absorb energy from another subsystem. Conversely, studying
the time varying case led to the canonical subsystem (Minkowski) satisfying the global energy and
66
momentum conservation, where the kinetic subsystems (Chu and Amperian) did not. The reason for
this stems from the definitions of the Minkowski formulation, where the electromagnetic and material
subsystem unified into one independent subsystem. That is, the canonical subsystem includes the
n o mat = TM + G
n o
Telec + Tmat + M ,
Gelec + G (147)
t t
where subscript M denotes the Minkowski electromagnetic subsystem. This, however, doesnt hold
true for the kinetic formulations, and is due to the separations of field and material contributions.
When modeling the moving system, both the Chu and Amperian formulation define field and ma-
terial contributions such that when only viewing the electromagnetic and mechanical subsystems,
the material contribution is still unaccounted for. That is, the kinetic formulations partition field
and material contributions such that when summing the respective electromagnetic and mechanical
contributions, there is still a remain partitioned material contribution unaccounted for. This is ob-
served in Figs 7 and 8, where the graphical difference between the two respective plots demonstrates
the material contribution of the modeled system in terms of the formulation used. Thus within each
n o mech = Tmat G
n o
Tk + Tmech + mat 6= 0,
Gk + G (148)
t t
demonstrates the necessary conditions for satisfying global conservations laws, where subscript k
denotes the respective kinetic formulations independently (either Chu or Amperian). Physically,
this reveals that however one partitions the overall system, it is relevant and necessary to account
for the material contribution when modeling the system in the time varying frame. Furthermore,
this demonstrates that one may partition the field and material contributions in any way as long
as each contribution sums to satisfy Maxwells equations, reasserting the view in [50]. However,
when dividing the material contributions within a given system, there are unique force expression
linked to specific energy expression, which satisfy electromagnetic theory. Within this Chapter, the
Chu and Amperian formulations were chosen, where both formulations have been shown (either
energy relations. Thus, it is satisfactory to say that addition of the field and material contributions
would validate each kinetic formulation. However, this is left to future research, where the study of
the overall system should reveal the material relations of each studied electromagnetic formulation.
67
This in turn will allow for a complete theoretical study of the partitioning of field and material
contributions, where the kinetics of light within materials can be fully analyzed.
In this Chapter, the electromagnetic force and power was studied with respect to three lead-
ing formulations of electrodynamics (Minkowski, Chu, and Amperian) for both time average and
time varying cases. The time average case yielded vacuum quantities for both the force and power
expressions, which resulted in identical relations for each formulation used. Thus, when modeling
the system in terms of any formulation, identical system dynamics were produced. This allowed
for a general understanding of the electromagnetic force and power produced for a moving slab
of material, but rendered nothing towards the understanding of the electromagnetic differences of
the formulations inside the material. The time varying results, however, demonstrated that the
electromagnetic force and power distributions rendered all subsystems in terms of both space and
time, which allowed for the study of the material subsystem. Within the analysis, the electromag-
netic and mechanical subsystems where modeled for each respective formulation. As was shown,
the Minkowski formulation, which included all relevant field and material terms, satisfied the global
energy and momentum conservation, where the Chu and Amperian formulations did not. This is
due to the partitioning of the material responses, of which, where not included in the global conser-
vation statement. However, due to the definitions within each formulation, addition of the material
contribution within the moving system should satisfy global conservation laws and demonstrate the
accurate partitioning of field and material contributions within a moving system. Thus, the meth-
ods shown serve to provide a deeper understanding into the kinetic subsystem, and provide the
68
.
6 Conclusion
This thesis serves to provide a consistent mathematical framework for effectively modeling contin-
uum electrodynamic momentum transfer to moving macroscopic media. This is based on classical
electromagnetic wave theory applied to Doppler shifted monochromatic waves propagating in mov-
ing linear, isotropic, and non-dispersive media. The presented results utilize a relativistic framework
for deriving the kinetic formulation of light within materials, which has been of interest and studied
for the past century. The methods demonstrated are shown to be consistent with valid physical prin-
ciples and mathematical truths, and validate both kinetic and canonical optical momentum transfer
to media. Thus, this theory is shown to be consistent with electromagnetic theory and is applied to
The theoretical work presented allows for a number of conceptual conclusions, of which are outline
here. This serves to build a better fundamental understanding of continuum electrodynamics and
Throughout this thesis, conservation principles are utilized to validate each physical system such
that global conservation, a fundamental tenet of physics, is satisfied. This analysis relied heavily on
the continuity equations, which serve to prove global conservation of energy and momentum, and
were derived in terms of the systems stress tensor, momentum density, power flux, and energy density.
By application with respect to each formulation, one can derive the dynamics of a system, where the
resulting forces and powers of each subsystem sum to zero. By employ electromagnetic force and
power definitions, one can study the energy and momentum in terms of each respective formulation.
Additionally, each formulation models one of two photon momenta. This was studied in Chapter 3,
were two formulations, Chu and Minkowski, were used to model both the Abraham and Minkowski
momenta, respectively. This study theoretically demonstrates the Abraham-Minkowski dilemma and
presents analytic solutions for both the kinetic and canonical forces. In addition, electromagnetic
69
wave theory was used, thereby producing both optical forces in terms of electromagnetic waves.
Recent research has recast the debate in terms of quantum mechanics due to the open ended nature of
the momentum dilemma [15]. However, the results from Chapter 3 have shown that electromagnetic
wave theory accurately demonstrates both Abraham and Minkowski momentum, and has illustrated
the moving media approach for this debate. These results have been published in Phys. Rev. A [1],
The kinetic formulation models the kinetic properties of a photon within a material. Quantum
mechanically, the kinetic formulation takes on the identity of a particle with some specified energy.
However, this model is irrelevant and renders no meaning towards the underlying electromagnetic
properties. Within electromagnetic theory, the kinetic momentum is a wave that interact with a
material such that it renders a center-of-mass translation with respect to an object. Additionally,
electromagnetic theory has multiple formulations that make some attempt in modeling the kinetic
subsystem by division of field from material quantities. To isolate the relevant relation, the La-
grangian is used, which derives the systems dynamics by use of a priori energy relation. Due to
the longevity of the photon momentum debate, the kinetic energy relations are well studied and
presented within the literature [35, 36]. Chapter 4 utilized the given energy relations to study two
kinetic formulation, where one can determine the kinetic subsystem by applying the Lagrangian and
relativistic principle of virtual power. This resulted in the Chu formulation as being the uniquely
defined kinetic formulation of light, disproving the Einstein-Laub formulation. Additionally, these
Over the past century, the kinetic subsystem has been studied from many differing perspec-
tives. During this time, the Balazs box thought experiment mathematically validated the kinetic
momentum of a photon, thereby validating the kinetic momentum density and initializing the kinetic
subsystem. Additionally, other researchers who have revisited the pioneering box thought experi-
ment have reshaped the optical momentum debate allowing for deeper understandings with respect
to each momentum model [12, 16]. However, due to the differing mathematical and physical interpre-
tations of each kinetic formulation, it remains unclear as to the true representation of electromagnetic
field relations within materials. In Chapter 5, we revisited the Balazs box thought experiment un-
der relativistic constraints to study the optical forces of each formulation. This demonstrated the
be consistent with physicals laws. As was demonstrated, the Minkowski formulation was shown to
account for both field and material subsystems, thereby proving total energy and momentum con-
servation for each studied case. This insinuates that the Minkowski formulation encompasses both
70
the canonical photon and material momenta. For the Chu and Amperian formulations, the kinetic
momentum was the only contribution modeled, and with respect to each formulation. The respective
kinetic models only derived the photon force, where the material subsystem was left unaccounted
for. This is due to the material subsystem being separated by each formulation into the respective
material contributions. However, modeling the material subsystem of each formulation will allow
for one to sum force and power densities, demonstrating global conservation, and allowing differing
electromagnetic formulations to be validated within the relativistic regime. This is a necessary con-
tribution in the optical momentum debate and is currently under progress towards publication in
Phys. Rev. A.
The results presented within this thesis demonstrate the study of relativistic electrodynamics
and kinetic subsystem. However, with any research, the results are incomplete. The theoretical
modeling of electromagnetic momentum could be applied indefinitely, with unique, creative, and
useful outcomes. During the course of my research, I have come across a number of interesting
topics which I think would aid in the understanding and current study of the kinetic momentum of
light and relativistic electrodynamics. In addition, some initial thoughts are given on how to begin
these topics.
Within this thesis, the Lagrangian approach and prescribed scalar and vector potentials are utilized
to formulate the Lagrangian density equation, of which is used to derive the equations of motion.
For completion, it would be wise to study the cited scalar and vector potentials by use of Greens
function. Here, Greens function would be used to derive expressions with respect to a point source,
H)
where the source is written in terms of a field vector (E, and material current density (Je , Jm ).
Greens functions are commonly used in finding field solutions in the presence of source distributions
for both antennas and radiation problems. Thus, expanding the field vectors into the cited scalar
and vector potential will allow for the validation of each term, thereby validating the Lagrangian
approach.
71
6.4 Addition of the material subsystem for time varying moving media
In Chapter 5, the study of time varying forces demonstrated the incomplete force expression for
electrodynamic formulations that separated field and material contributions. This partitioning of
material responses is a necessary requirement for understanding the total system. For completion,
it would be wise to study the formulation of the material responses. This would enable a complete
understanding of both field and material systems and would allow for the derivation of the kinetic
subsystem. Current research is utilizing these concepts to validate the Chu formulation as the kinetic
subsystem.
The topic of negative index materials is a recent and growing topic in optics. Having unique
properties, negative index materials induce interesting relations such as reversing wave vectors and
power flows within the material, as well as producing reversed Doppler shifts and reversed Cerenkov
radiation effects. Applications of relativistic electrodynamics in modeling the energy and momentum
would provide an interesting perspective of how light interacts with the causal negative index ma-
terials. This would allow for deeper understanding of the division of field an material contributions
within moving electromagnetic studies, and expand the growing body of knowledge of within the
72
Appendices
The Lorentz transformation is used to transform the material contributions from the moving frame
to the stationary frame. In addition, Minkowskis postulate utilized, which states the macroscopic
Maxwells equations are Lorentz-covariant, to transform the various formulations between the moving
Consider a moving, lossless, nondispersive, isotropic medium. To illustrate this, the constitutive
0 = 0 E0 and H
relations D 0 = B0 /0 are moving with velocity v = zv with respect to the stationary
c0 I 0
C 1
= L 6 L6 (149)
0 1
c 0 I
account for the Lorentz transformation in dyadic notation. The general form of the constitutive
relations
cD E
= C (150)
H cB
P L
C = (151)
M
Q
M,
L,
is the constitutive relations matrix, c = 3 108 m/s is the velocity of light, and P, and Q
are all 3 3 matrices involving material parameters. Here, the field relations E and cB form a
and H
four-dimensional space tensor such that they are covariant. Similarly, the field vectors cD
form a covariant four-dimensional space tensor by relation. Applying this to moving media with
73
velocity v = zv, we find
p 0 0 0 l 0
0 p 0 l 0 0
1 0 0 pz 0 0 0
C = 0 (152)
c
0 l 0 q 0 0
l 0 0 0 q 0
0 0 0 0 0 qz
where
p = 2 (n2 2 ) (153a)
q = 2 (1 n2 2 ) (153b)
l = 2 (n2 1) (153c)
pz = 2 n2 (1 2 ) (153d)
qz = 1. (153e)
The constitutive relations demonstrate bianisotropic media properties in the stationary frame.
Bianisotropic media provides a cross coupling between the electric and magnetic fields such that
they become both polarized and magnetized when placed in an electric or magnetic field [18, 51].
When an incident beam propagates within a moving, lossless, non-dispersive, isotropic medium, it
yields both polarization and magnetization material properties in the stationary frame.
Next, we use the kDB system [18, 27] and the constitutive relations to derive the wave vector
Here, we transform the stationary constitutive relations to utilize the constant orthogonality
k.
and B.
property of wave field vectors D We use the EH constitutive relations
D E
= CEH (154)
B
H
where
1 1
1 P L Q M LQ
CEH = = (155)
c
Q 1 M
1
Q
to formulate the needed field vectors. As is shown, the EH constitutive relations are written in terms
74
of the Eq. (151), where CEH is the EH constitutive matrix. When transforming the bianisotropic
0 0 0 0
0 0
0 0
0 0 z 0 0 0
CEH =
(156)
0 0 0 0
0 0 0 0
0 0 0 0 0 z
where
c2 (1 2 )0
= (157a)
n2 2
2 0
c (1 2 )
= (157b)
n2 2
c(n2 1)
= (157c)
n2 2
1
z = (157d)
0
1
z = . (157e)
0
k
T1
= T (158a)
k
= T
T1 (158b)
k = T T1 (158c)
k = T T1 (158d)
75
where
sin cos 0
T =
cos cos cos sin sin
(159a)
sin cos sin sin cos
sin cos cos sin cos
1
T =
(159b)
cos cos sin sin sin
0 sin cos
such that
E = k +
k D k
k B (160a)
k
H k + k D
= k B k. (160b)
k
k , ,
Here, the , and make up the DB constitutive matrix [18], and is used for transforms
0 0
k
= 2 2
0 cos + z sin ( z ) sin cos
(161a)
0 ( z ) sin cos cos2 + z sin2
0 0
k = 2 2
0 cos + z sin ( z ) sin cos
(161b)
0 ( z ) sin cos cos2 + z sin2
0 cos cos
k = k
=
cos 0 0
(161c)
cos 0 0
76
or D.
After simplification, we can solve the system in terms of B Here, we choose the D
field vector
2 2
D1 0 cos + z sin B1
u cos = . (163)
D2 0 B2
(u cos )2
1 =0 (164a)
( cos2 + z sin2 )
(u cos )2
1 =0 (164b)
( cos2 + z sin2 )
( kz )2
kz2 + kx2 + ky2 =0 (165a)
z z
( kz )2
kz2 + kx2 + ky2 =0 (165b)
z z
n
k = z (166a)
n 1 c
n+
k = z (166b)
n + 1 c
where Eq. (166a) corresponds to a wave propagating in the negative z direction while Eq. (166b)
refers to a wave propagating in the positive z direction. This result is attained for the duel case by
B Field Relations
Consider a plane wave normally incident upon a moving magneto-dielectric halfspace. Here, the
material substrate is moving such that v = zv where the plane wave is propagating in the
z
direction. Thus, the boundary for the material is defined at z = vt. Additionally, let the electric
77
field be linearly polarized in the x
direction such that,
E (
r, t) = E0 ei(k z t)
x (167a)
1 E0 n i(k z t)
B (
r, t) = [k E ] = y e (167b)
c n 1
where k is given in Eq.(166a), and the subscript denotes waves traveling in the negative z
(
direction. To calculate field vectors cD (
r, t) and H r, t), we make use of the constitutive relations
cD
x p 0 0 0 l 0 Ex
0 0 p 0 l 0 0 0
0 0 0 pz 0 0 0 0
= 1 (168)
c0
0
0
l 0 q 0 0 0
H l 0 0 0 q 0 cBy
y
0 0 0 0 0 0 qz 0
to yield
1
cDx = (p Ex l cBy ) (169a)
c0
1
Hy = (l Ex + q cBy ) (169b)
c0
ck Ex
where l, p, and q are defined in Eq.(153). Here, we make use of the relation cBy = to rewrite
1 cki 1 n(n )
cDx = 0
(p l )Ex = Ex (170a)
c c0 0r (1 n)
1 ck n
Hy = 0
(l + q )Ex = Ex . (170b)
c c0 0r
(
Using this, we find cD (
r, t) and H r, t) as
( E0 n(n ) i(k z t)
D r, t) = x
e (171a)
c2 0 1 n
( n
H r, t) = y 0 E0 ei(k z t) (171b)
c
78
Similarly, using the constitutive matrix and defining k+ with Eq.(166b), the wave vectors traveling
E+ (
r, t) = E0 Rei(k+ z+ t)
x (172a)
E0 n +
B+ (
r, t) = y Rei(k+ z+ t) (172b)
c n + 1
+ ( E0 n(n + ) i(k+ z+ t)
D r, t) = x 2 Re (172c)
c 1 + n
+ ( E0
H r, t) = yn Rei(k+ z+ t) . (172d)
c
For the specific dielectric case, we take 0 0 , for the above wave relations.
To evaluate the case for the linearly polarized magnetic wave, we take the dual of the case presented
cB H
= Cdual . (173)
E cD
Now, using duality, the constitutive relations in the moving frame transform to B0 = 0 H
and
E0 = D
0 /0 . Using this, we transform these relations to the stationary frame such that
c0 I 0
Cdual 1
=L 6 L6 (174)
0 1
c0I
to yield
p 0 0 0 l 0
0 p 0 l 0 0
0 0 pz 0 0 0
1
Cdual = 0 (175)
c 0
l 0 q 0 0
l 0 0 0 q 0
0 0 0 0 0 qz
79
where
p = 2 (n2 2 ) (176a)
q = 2 (1 n2 2 ) (176b)
l = 2 (n2 1) (176c)
pz = 2 n2 (1 2 ) (176d)
qz = 1. (176e)
Here, we consider a linear magnetic wave propagating within a dielectric halfspace normally incident
upon a perfect magnetic conductor (PMC). The linear magnetic wave is polarized in the y direction
such that
(
H r, t) = yH0 ei(k z t) (177a)
( 1
D r, t) = k H
i
H0 n i(ki zi t)
= x e , (177b)
c n 1
0 p 0 0 0 l 0 0
Hy
cB 0 p 0 l 0 0
y
0 0 0 pz 0 0 0 0
= 1 (178)
c0
Ex 0 l 0 q 0 0 cDx
0 l 0 0 0 q 0 0
0 0 0 0 0 0 qz 0
such that
1
cBy = (pHy lcDx ) (179a)
c0
1
Ex = (lHy qcDx ). (179b)
c0
80
cki Hy
Here, we make use of the relation cDy = to find
1 k c 1 n(n )
cBiy = (p l )Hy = Hy (180a)
c0 c0 0r 1 n
1 ki c n
Eix = (l + q )Hy = Hy (180b)
c0 c0 0r
H0 n(n ) i(k z t)
B (
r, t) = y e (181a)
c2 0 0r (1 n)
H0 n i(k z t)
E (
r, t) =
x e . (181b)
c0 0r
Using the the constitutive matrix and defining kr with Eq.(166b), we find the reflected waves to be
+ (
H r, t) = y H0 Rei(k+ z+ t)
(182a)
+ ( H0 n +
D r, t) =
x Rei(k+ z+ t) (182b)
c n + 1
H0 n i(k+ z+ t)
E+ (
r, t) =
x Re (182c)
c0 0r
H0 n(n + )
B+ (
r, t) =
y 2 Rei(k+ z+ t) . (182d)
c 0 0r (1 + n)
Lagranges equations derive the equations of motion for any system given the systems energy
relations. Here, we derive the equations which describe the kinetic subsystem. This is done by
2 2
A C
L(e , A, = 0
m , C) 1
e C
0
m +
1
A
2 0 t 2 0 t
e e + Je A + m m + Jm C (183)
d L L d L
= . (184)
dt x j xj dXK (xj /XK )
81
For the electric scalar potential, we find the Lagrangian with respect to e by
d L L d L
= (185)
dt (e /t) e dzj e,j
to yield
A
d 1
0 = e 0 e,j C (1)
dzj 0 t
d
= e + 0 E
dzj
=
e + 0 E
0 E = e (186)
Similarly, we find the Lagrangian with respect to the magnetic scalar potential m by
d L L d L
= (187)
dt (m /t) m dzj m,j
yielding
C
d 1
0 = m 0 m,j + A (1)
dzj 0 t
d
= m 0 H
dzj
=
m 0 H
0 H = m (188)
d L L d L
= . (189)
dt ( Ai /t) Ai dz
j ( Am /zl )
to yield
A Am C
d 1 d 1 kji
0 e C (1) = Jei 0 m + klm ( )
dt 0 t dzj 0 zl t 0
d d
0 E = Jei (kji ) H
dt dzj
E
0 = Je H
t
H 0 E = Je (190)
t
82
where the permutation symbol klm is used. Lastly, the Lagrangian in terms of vector potential C
is found by
d L L d L
= . (191)
dt ( Ci /t) Ci dzj ( Cm /zl )
yielding
C Cm A
d 1 d 1 kji
0 m + A (1) = Jmi 0 e klm ( )
dt 0 t dzj 0 zl t 0
d d
0 H = Jmi + (kji ) E
dt dzj
H
0 = Jm + E
t
E + 0 H = Jm . (192)
t
As can be seen, the Maxwell-Chu formulation is derived where the sum of Eqs.(186-192) are identical
to Eqs.(12).
This sections illustrates the techniques used in deriving the stress tensor, momentum density,
energy density, and Poynting power of an electrodynamic formulation. Specifically, the expression
for the Maxwell-Chu equations, Eqs.(12), are derived for the electromagnetic subsystem.
To derive the energy density and Poynting power, we perform the standard manipulation in
H
H E = 0 Jm
t
E
E H = 0
+ Je
t
( E)
E
H
= H 0 H + Jm E 0 E + Je
H
t t
H
+H 0 H + E 0 E = E Je H
Jm
E
t t
83
Y ) = Y ( X)
where (X X
( Y ) has been used. Additionally, 2 A(t)
t A(t) = 2A(t) t
A(t)
implies A(t) = 1 2
t 2 t A(t) . Thus,
0 (E + 0 (H
h i
Je + H
Jm H E) H)
=E = E (193)
t 2 2
= S W
t
where
S = H)
(E (194a)
h 0 i
0
W = (E E) + (H H) . (194b)
2 2
The stress tensor and momentum density are found by employing vector field definitions such
that,
H = 0 E + Je 0 H
t
+ E = 0 H Jm 0 E
t
= m H
+ 0 H
= e E
+ 0 E
which gives
0 H)
( H + ( E
0 E)
+ H
( 0 H) + ( 0 E)
E
E
= (0 (0 H + Jm ) 0 E
+ Je ) 0 H + m H
+ e E.
t t
84
After rearranging,
0 [( H) H + ( H)
H]
+ 0 [( E)
E
+ ( E)
E]
(195a)
H
+ 0 0 E 0 E 0 H
(195b)
t t
= + Je 0 H
e E + m H Jm 0 E
(195c)
= (A A)
h i
A + ( A)
[( A) A] I AA
2
we find
h i
E
0 [( E) + ( E)
E]
=
0 I 0 E
E)
(E E
2
h i
H
0 [( H) + ( H)
H]
=
0 I 0 H
H)
(H H .
2
Addition of these two results render the stress tensor, redefining Eq.(195a). Thus,
H
0 [( H) + ( H)
H] + 0 [( E)
E + ( E)
E]
h 0 i
0 E 0 H
H
= (E E) + (H H) I 0 E
2 2
H
0 0 E 0 H
0 E = 0 0
H
E E H
t t t t
= 0 0 E H + E H
t t
= 0 0 [E H]
t
Now,
f = e E
+ Je 0 H
+ m H
Jm 0 E
h 0 0 0 [E
i
0 E 0 H
H H]
= (E E) + (H H) I 0 E
2 2 t
G
= T
t
85
where
+ 0 (H
T I 0 E
0 E) H) E 0 H
H
= (E (196a)
2 2
=
G H),
0 0 (E (196b)
demonstrating the electromagnetic stress tensor and momentum density for the Chu formulation.
86
References
[1] C. J. Sheppard and B. A. Kemp. Optical pressure deduced from energy relations within rela-
[3] Albert Einstein et al. On the electrodynamics of moving bodies. Annalen der Physik,
17(10):891921, 1905.
[4] J. H. Poynting. Tangential stress of light obliquely incident on absorbing surface. Phil. Mag,
24:156, 1905.
K
orpern. Nachr. Ges. Wiss. G
ottingen, 1:53, 1908.
[6] M. Abraham. Zur elektrodynamik der bewegter Korper. Rend. Pal., 28:1, 1909.
[7] A. Ashkin. Optical trapping and manipulation of neutral particles using lasers. Proc. Natl.
[8] A. Ashkin. History of optical trapping and manipulation of small-neutral particle, atoms, and
[9] A. Ashkin and J. M. Dziedzic. Optical trapping and manipulation of viruses and bacteria.
[10] Kishan Dholakia and Pavel Zemnek. Colloquium: Gripped by light: Optical binding. Rev. of
[12] S. M. Barnett. Resolution of the Abraham-Minkowski dilemma. Phys. Rev. Lett., 104:070401,
February 2010.
[13] B. A. Kemp and T. M. Grzegorczyk. The observable pressure of light in dielectric fluids. Opt.
87
[16] B. A. Kemp. Resolution of the Abraham-Minkowski debate: Implications for the electromag-
[17] Robert M Fano, Lan Jen Chu, Richard B Adler, et al. Electromagnetic fields, energy, and
[18] J. A. Kong. Electromagnetic Wave Theory. EMW Publishing, Cambridge, MA, 2005.
[19] NL Balazs. The energy-momentum tensor of the electromagnetic field inside matter. Physical
[20] R. V. Jones and J. C. S. Richards. The pressure of radiation in a refracting medium. Proc. R.
[21] R. V. Jones and B. Leslie. The measurement of optical radiation pressure in dispersive media.
[22] M. Mansuripur. Radiation pressure on submerged mirrors: Implications for the momentum of
[23] P. Daly and H. Gruenberg. Energy relations for plane waves reflected from moving media.
Haus. Radiation pressure of light pulses and conservation of linear momentum in dispersive
[25] R. Loudon. Radiation pressure and momentum in dielectrics. Fortschr. Phys, 52:1134, 2004.
[26] P. Penfield and H. A. Haus. Electrodynamics of Moving Media. M.I.T. Press, Cambridge, MA,
1967.
[27] Jin Au Kong. Theory of electromagnetic waves. New York, Wiley-Interscience, 1975. 348 p.,
1, 1975.
NanoScience+ Engineering, pages 845803845803. International Society for Optics and Pho-
tonics, 2012.
[29] Kevin J Webb. Dependence of the radiation pressure on the background refractive index.
88
[30] M. Mansuripur. Deducing radiation pressure on a submerged mirror from the doppler shift.
[31] L. Brillouin. Wave Propagation and Group Velocity. Academic Press, New York, 1960.
[34] R. Loudon, L. Allen, and D. F Nelson. Propagation of electromagnetic energy and momentum
[37] Masud Mansuripur. Optical manipulation: Momentum exchange effect. Nature Photonics,
7(10):765766, 2013.
[38] D. F. Nelson. Electric, Optic, and Acoustic Interactions in Dielectrics. John Wiley and Sons,
[39] D. F. Nelson. Momentum, pseudomomentum, and wave momentum: Toward resolving the
[41] Changbiao Wang. Plane wave in a moving medium and resolution of the Abraham-Minkowski
[42] Changbiao Wang. Can the Abraham light momentum and energy in a medium constitute a
[43] Changbiao Wang. Self-consistent theory for a plane wave in a moving medium and light-
89
[44] Douglas Singleton. Magnetic charge as a hidden gauge symmetry. International Journal of
[45] Chen-To Tai. Direct integration of field equations. Progress In Electromagnetics Research,
28:339359, 2000.
[46] Robert Schulmann, Anne J Kox, Michel Janssen, and Jozsef Illy. The collected papers of Albert
[47] Masud Mansuripur, Armis R Zakharian, and Ewan M Wright. Electromagnetic-force distribu-
[48] Brandon A Kemp. The kinetic subsystem of light and its role in optical manipulation. In
SPIE NanoScience+ Engineering, pages 88100J88100J. International Society for Optics and
Photonics, 2013.
[49] Masud Mansuripur. Resolution of the Abraham - Minkowski controversy. Opt. Commun.,
283(10):1997, 2010.
[51] Jin Au Kong. Theorems of bianisotropic media. Proceedings of the IEEE, 60(9):10361046,
1972.
90