You are on page 1of 12

Contents

1 Preliminaries 2
1.1 Current Density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Lorentz Force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 Maxwell’s Equations 3
2.1 Maxwell’s Equations in (2+1) Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Longitudinal and Transverse Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

3 Electromagnetic Fields in Matter 6


3.1 Dielectrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.1.1 Macroscopic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.1.2 Microscopic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.2 Magnetic Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.2.1 Magnetization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.2.2 Field Produced by Magnetized Matter . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.2.3 Example: Uniformly Magnetized Sphere . . . . . . . . . . . . . . . . . . . . . . . . 10
3.2.4 Total Magnetic Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.2.5 Simple Magnetic Matter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1
1 Preliminaries
1.1 Current Density
Let n̂ be the local unit normal to an element of surface dS, such that dS = n̂ dS (see (a) in the Figure
below). We define a current density j so that dI = j · dS is the rate at which charge passes through dS.
The total current that passes through a surface S is then given by
Z
I= j · dS
S

When a velocity field v characterizes the motion of a charge density ρ, the current density is given by
j = ρv
If the charge is entirely confined to a two-dimensional surface, we define instead a surface current
density in terms of the surface charge density σ:
K = σv
The current which flows past a curve C on a surface (see (b) in the Figure below) can then be expressed
in terms of the local surface normal n̂ using
Z Z
I= (K × n̂) · dl = K · (n̂ × dl)
C C

1.2 Lorentz Force


(Following Reitz and Milford, Foundations of Electromagnetic Theory, 1st ed) In the absence of an electric
field, the magnetic field B is defined as the vector which satisfies
F = qv × B
for all velocities. It is important to note that a single measurement of the force on a charge with
a particular velocity is not sufficient to determine B (see Mathematics notes on inverting the vector
product). However, if two measurements are made for two mutually perpendicular velocities, v1 and v2 ,
we can obtain a unique expression for B.
Using the standard result from vector analysis, we obtain for v1
F1 × v1
B= + k1 v1
qv12
and for v2
F2 × v 2
B= + k2 v2
qv22
where k1 and k2 are arbitrary scalars. In order to eliminate of these constants we take the scalar product
of both of the above expressions with v1 and equate them, to yield
v1 · (F1 × v1 ) v1 · (F2 × v2 )
+ k1 v1 · v1 = + k2 v1 · v2
qv12 qv22

2
The first term on the LHS vanishes identically, and the second term on the RHS is also zero since v1 and
v2 are mutually perpendicular. Rearranging then gives
v1 · (F2 × v2 )
k1 v12 =
qv22
Substituting for k1 thus obtained in the first of the original expressions for B then yields
F1 × v 1 v1 · (F2 × v2 )
B= + v1
qv12 qv12 v22
which explicitly shows that two distinct measurements are sufficient to uniquely determine B.

2 Maxwell’s Equations
We begin with the inhomogeneous form of Maxwell’s equations, from which it is apparent that all elec-
tromagnetic fields are ultimately attributable to charges and currents:

∇·E=ρ
 
1 ∂E
∇×B= J+
c ∂t

∇·B=0
1 ∂B
∇×E=−
c ∂t
The first two of these equations are consistent with the observation of charge conservation, expressed by
the continuity equation:
∂ρ
+∇·J=0
∂t
If we further postulate that ρ and J make up a contravariant four-vector field j µ = (cρ, J), then the
continuity equation takes the Lorentz-covariant form

∂µ j µ = 0

where ∂µ = (c−1 ∂/∂t, ∂/∂x, ∂/∂y, ∂/∂z) and ∂ µ = (c−1 ∂/∂t, −∂/∂x, −∂/∂y, −∂/∂z).
We now introduce a scalar potential ϕ and a vector potential A that satisfy the last two Maxwell
equations identically with
1 ∂A
E = −∇ϕ −
c ∂t
B=∇×A

We can postulate that these potentials also make up a four-vector Aµ = (ϕ, A), or Aµ = (ϕ, −A), so that
the above two equations can be encapsulated as

∂ µ Aµ = 0

If we also introduce the notation

E ≡ (E 1 , E 2 , E 3 ) ≡ (−E1 , −E2 , −E3 ) ≡ (Ex , Ey , Ez )

B ≡ (B 1 , B 2 , B 3 ) ≡ (−B1 , −B2 , −B3 ) ≡ (Bx , By , Bz )

then we can introduce the (antisymmetric) field-strength tensor

E1 E2 E3
   
0 0 Ex Ey Ez
−E 1 0 −B 3 B 2
 −Ex 0 −Bz By 
Fµν ≡ ∂µ Aν − ∂ν Aµ = −E 2 B3
= 
0 −B1  −Ey Bz 0 −Bx 
−E 3 −B2 B1 0 −Ez −By Bx 0

3
where µ, ν = 0, 1, 2, 3, or alternatively
−E 1 −E 2 −E 3
   
0 0 −Ex −Ey −Ez
1
E 0 −B 3 B2  Ex 0 −Bz By 
F µν ≡ ∂ µ Aν − ∂ ν Aµ = 

= 
E 2 B3 0 −B 1  Ey Bz 0 −Bx 
E3 −B 2 B1 0 Ez −By Bx 0
From the above, we find that the components of the electric field are given by
Ex ≡ E 1 = ∂0 A1 − ∂1 A0 = ∂ 1 A0 − ∂ 0 A1 




Ey ≡ E 2 = ∂0 A2 − ∂2 A0 = ∂ 2 A0 − ∂ 0 A2 =⇒ E i = ∂0 Ai − ∂i A0 = ∂ i A0 − ∂ 0 Ai (i = 1, 2, 3)



Ez ≡ E 3 = ∂0 A3 − ∂3 A0 = ∂ 3 A0 − ∂ 0 A3

while the components of the magnetic field are given by


Bx ≡ B 1 = ∂3 A2 − ∂2 A3 = ∂ 3 A2 − ∂ 2 A3 




2 1 3
By ≡ B = ∂1 A3 − ∂3 A1 = ∂ A − ∂ A 3 1
=⇒ B i = −ϵijk ∂j Ak = −ϵijk ∂ j Ak (i, j, k = 1, 2, 3)



Bz ≡ B 3 = ∂2 A1 − ∂1 A2 = ∂ 2 A1 − ∂ 1 A2

The first two of Maxwell’s equations above can now be expressed in compact form using F µν :

∂µ F µν =
c
To see this, we first note that ∂µ = ((1/c)∂/∂t, ∂/∂x, ∂/∂y, ∂/∂z) and J ν = (ρc, J). The LHS of the first
Maxwell equation results from operating with each element of the row vector ∂µ on the corresponding
elements of the first column of F µν and summing the result, while the RHS results from multiplying the
first element of the row vector J ν by 1/c. The LHS of the second Maxwell equation results from operating
with each element of the row vector ∂µ on the second, third and fourth columns of F µν with the resulting
sum for each column forming the components of a vector, while the RHS results from multiplying the
second, third and fourth elements of the row vector J ν by 1/c to form another vector.
The second two of Maxwell’s equations can be obtained from the Bianchi identity
∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0
where λ, µ, ν must be different from one another for LHS to be nonzero. The third Maxwell equation is
obtained from λ = 1, 2, 3, while the fourth Maxwell equation is obtained from λ = 0.

2.1 Maxwell’s Equations in (2+1) Dimensions


In (2+1) dimensions the electric field is a 2D vector:
E ≡ (E 1 , E 2 ) ≡ (−E1 , −E2 ) ≡ (Ex , Ey )
while the magnetic field is a pseudoscalar, B ≡ Bz . The field-strength tensor is then
E1 E2
   
0 0 Ex Ey
Fµν ≡ ∂µ Aν − ∂ν Aµ = −E 1 0 −B  = −Ex 0 −B 
−E 2 B 0 −Ey B 0
or alternatively
−E 1 −E 2
   
0 0 −Ex −Ey
F µν ≡ ∂ µ Aν − ∂ ν Aµ =  E 1 0 −B  = Ex 0 −B 
E2 B 0 Ey B 0
The components of the electric field are given by

Ex ≡ E 1 = ∂0 A1 − ∂1 A0 = ∂ 1 A0 − ∂ 0 A1 
=⇒ E i = ∂0 Ai − ∂i A0 = ∂ i A0 − ∂ 0 Ai (i = 1, 2)
2 2 0 0 2
Ey ≡ E = ∂0 A2 − ∂2 A0 = ∂ A − ∂ A
while the magnetic field is given by
B = ∂2 A1 − ∂1 A2 = ∂ 2 A1 − ∂ 1 A2 =⇒ B = −ϵij ∂i Aj = −ϵij ∂ i Aj (i, j = 1, 2) (1)

4
2.2 Longitudinal and Transverse Fields
Maxwell’s equations of classical electromagnetism can be expressed in modified Heaviside-Lorentz units
as follows:
1 ∂B(r, t)
∇ · E(r, t) = ρtot (r, t), ∇ × E(r, t) + =0
c ∂t
1 ∂E(r, t) 1
∇ · B(r, t) = 0, ∇ × B(r, t) − = Jtot (r, t)
c ∂t c
where E(r, t) and B(r, t) are the self-consistent local fields, determined in part by the charges and currents
they induce, and ρtot (r, t) and Jtot (r, t) are the total charge and current densities, which are comprised
of the external charge and current densities of the sources that give rise to the external fields and the
charge and current densities induced in the system by these sources/fields, i.e.

ρtot (r, t) ≡ ρex (r, t) + ρ(r, t), Jtot (r, t) ≡ Jex (r, t) + J(r, t)

If we now introduce scalar and vector potentials via


1 ∂A(r, t)
E(r, t) = −∇ϕ(r, t) − , B(r, t) = ∇ × A(r, t)
c ∂t
where ϕ(r, t) and A(r, t) include contributions from both the external field and the polarization field
induced by the latter, then substituting these potentials into Maxwell’s equations results in two equations
becoming identically zero, while the other two can be written as (suppressing the spatial and time
dependence from now on for clarity):
1 ∂
∇2 ϕ + (∇ · A) = −ρtot
c ∂t
1 ∂2A 1
 
2 ∂ϕ 1
∇ A − ∇(∇ · A) − 2 2 − ∇ = − Jtot
c ∂t c ∂t c
To simplify the above, we can introduce the Coulomb gauge, ∇ · A = 0, to yield

∇2 ϕ = −ρtot

1 ∂2A 1
 
2 ∂ϕ 1
∇ A− 2 2 − ∇ = − Jtot
c ∂t c ∂t c
The first equation is simply Poisson’s equation, while the second is a wave equation for A - the electromag-
netic field has therefore been split into an electrostatic potential (described by ϕ) and an electromagnetic
wave (described by A).
A fundamental result of vector calculus says that any vector field can be split into longitudinal and
transverse components, so that we have, for example:

E = ET + EL

where by definition
∇ × EL = 0 and ∇ · ET = 0
tot
and similarly, B = BT + BL , J = Jtot
T + Jtot
L and A = AT + Atot
L . Applying the Coulomb gauge
condition to the latter gives
∇ · A = ∇ · (AT + AL ) = 0
and because ∇ · AT = 0 by definition, we must also have ∇ · AL = 0, so that in the Coulomb gauge
A only has a transverse component (hence the alternative name, transverse gauge). Note that we could
have chosen an alternative gauge, in which case A would have had both longitudinal and transverse
components.
Decomposing the current density allows us to use the continuity equation to obtain
∂ρtot
− = ∇ · Jtot = ∇ · (Jtot tot tot
T + JL ) = ∇ · J L
∂t

5
Differentiating Poisson’s equation with respect to time gives

∂ρtot
 
∂ ∂ ∂ϕ
(∇2 ϕ) = (∇ · ∇ϕ) = ∇ · ∇ =−
∂t ∂t ∂t ∂t

and substituting the continuity equation obtained above then yields


 
∂ϕ
∇ = Jtot
L
∂t

If we then substitute this into the wave equation for A, we obtain


1 ∂AT 1
∇ 2 AT − = − Jtot
c2 ∂t c T
Hence, we now have two decoupled equations - one in terms of ϕ and Jtot
L , and the other in terms of A
and Jtot
T .

3 Electromagnetic Fields in Matter


3.1 Dielectrics
3.1.1 Macroscopic Theory
A dielectric is a medium that cannot completely screen a static, external, macroscopic electric field from
its interior. This is due to chemical bonding and other quantum mechanical effects which constrain
the rearrangement of its internal charge density. The same constraints are responsible for the fact that
dielectrics do not conduct (or poorly conduct) electric current. A dielectric responds to an external field
Eext (r) by distorting its ground state charge density to produce a field EP (r). The total macroscopic
electric field that enters Maxwell’s theory is the sum of these two fields, and is non-zero both inside and
outside the volume of the dielectric, i.e.

E(r) = Eext (r) + EP (r) ̸= 0

The rearrangement of the internal charge of a dielectric resulting from the application of an external
field is characterized by the polarization P(r), and it is traditionally defined as the electric dipole moment
per unit volume, so that the integral of the polarization over the volume of a dielectric is equal to the
total dipole moment of the dielectric, i.e.
Z
d3 r P(r) = p
V

The source of the electric field EP (r) arising from polarization of the medium is referred to as bound
charge or polarization charge, with volume density ρP (r) = −∇ · P(r). On the other hand, the source of
the external electric field Eext (r) is referred to as free charge, with volume density ρf (r). Examples of free
charge include the charge on the surface of capacitor plates and point charges placed inside or outside
the body of a dielectric. The total charge density that enters Maxwell’s theory is the sum of the free and
bound charge densities, i.e.
ρ(r) = ρf (r) + ρP (r)
Applying Gauss’s law then yields

ϵ0 ∇ · E(r) = ρ(r) = ρf (r) + ρP (r) = −∇ · P(r) + ρf (r)

We can define a new vector field D(r), traditionally known as the electric displacement, as follows:

D(r) ≡ ε0 E(r) + P(r)

while the fundamental electrostatic condition remains valid in matter, i.e.

∇×E=0

6
The above three equations therefore imply that

∇ · D = ρf

and
∇×D=∇×P
The above set of equations is insufficient to determine all the unknown quantities, and must be
supplemented by a so-called constitutive relation that relates P(r) to E(r) (or equivalently, D(r) to
E(r)) obtained from experiment, theory or phenomenology. For the vast majority of systems, which are
unpolarized in the absence of a field but acquire a uniform macroscopic polarization in the presence of a
uniform external electric field, the general rule revealed by experiment is
(2)
Pi = ε0 χij Ej + ε0 χijk Ej Ek + . . .
(2)
where the tensor nature of the constants χij and χij allows for the possibility that P is not parallel to
E, as is the case in spatially anisotropic matter. The second- and higher-order terms in the expansion
allow for the possibility that the polarization depends non-linearly on the field, which occurs in all matter
when the electric field strength is large enough.
If a dielectric is linear, it can be described by just the first term on the right-hand side of the expansion;
if it is also isotropic, then the tensor χij can be replaced by a scalar and the constitutive relation written
simply as
P = ε0 χE
The constant χ is called the electric susceptibility. We also introduce the (dimensionless) relative permit-
tivity εr ≡ 1 + χ (sometimes also known as the dielectric constant, κ) via

P = ε0 (εr − 1)E

In a linear, isotropic medium the electric displacement D defined earlier can therefore be expressed as

D = εr ε0 E = ε0 (1 + χ)E

3.1.2 Microscopic Theory


So far, we have considered only large-scale phenomena in dielectrics and a model involving continuous
distributions of charge, with macroscopic electric fields arising from bound and free charge densities.
However, matter is in reality discrete, and the microscopic electric field, or force per unit charge, inside
a dielectric material will therefore fluctuate significantly from point to point within a dielectric.
To give some physical significance to the macroscopic field within a dielectric we imagine hollowing
out a spherical cavity within the material which is macroscopically small but large enough for a point
within it to be considered outside the dielectric, so that we obtain a force per unit charge that does not
fluctuate like the microscopic field. The polarization charges on the inner surface of the sphere produce
an electric field at its centre given by (see Duffin, p.296):
Z π
−P cos2 θ P
d(cos θ) =
0 2ε0 3ε0
where θ is the angle between the polarization vector P and a point on the surface of the sphere. The
total electric field inside the cavity will therefore be given, in terms of the macroscopic field E, by
P
E+
3ε0
We now consider the electric field experienced by a single atom in a dielectric, which we call the local
electric field, Elocal . We then define the polarizability α of such an atom as the proportionality constant
between the local electric field and the dipole moment p induced in the body by that field, i.e.

p = αElocal

The polarization will then be given in terms of the number of dipoles per unit volume, n, by

P = nαElocal

7
However, we already have an expression for the polarization in terms of the dielectric constant and
macroscopic field, i.e. P = ε0 (εr − 1)E, and eliminating P therefore yields

nα Elocal
κ=1+
ε0 E
The question now is how to determine the local field. Lorentz suggested imagining a macroscopically
small sphere centred on the atom in question - all the material outside the sphere then contributes the
field E + P/3ϵ0 calculated earlier, while all the dipoles within the sphere must be treated individually and
their effect added to the latter field. For a material with a cubic lattice, he showed that the contribution
of the dipoles within the sphere would be zero, and hence
P
Elocal = E +
3ε0

from which, using P = ε0 (εr − 1)E, we obtain


 
nα εr − 1
κ=1+ 1+
ε0 3

which can be rearranged to yield


εr − 1 nα
=
εr + 2 3ε0
This is known as the Clausius-Mossotti relation, and provides a valuable link between macroscopic and
microscopic theories via the dielectric constant and polarizability, respectively.

3.2 Magnetic Materials


3.2.1 Magnetization
All matter responds to a static external magnetic field, Bext , by producing a magnetic field of its own,
BM . The total magnetic field, both inside and outside the matter, is the sum of the external and induced
field, i.e.
B(r) = Bext (r) + BM (r)
Far outside the matter, BM is dipolar and thus may be characterized by a macroscopic magnetic dipole
moment m, the origin of which is fundamentally quantum-mechanical. The magnetic moment of a
paramagnet points parallel to Bext , while the magnetic moment of a diamagnet points anti-parallel to
Bext . For most magnets, the induced moment vanishes when the external field is removed, but the
exception to this rule is a ferromagnet, which is the special case of a paramagnet where m remains non-
zero when Bext → 0. A superconductor is a special case of a diamagnet where the total field B is zero
inside its volume.

All magnetic phenomena are due to electric charges in motion. At the atomic scale, a piece of magnetic
material contains microscopic currents due to electrons orbiting around nuclei and spinning on their own
axes. On a macroscopic scale, these current loops are so small that we can treat them as tiny magnetic
dipoles. Usually, these dipoles cancel each other out due to the random orientation of the atoms, but
when the material is placed in an external magnetic field, there is a net alignment of the dipoles and
the material becomes magnetically polarized, or magnetized. For a magnetized material, we define the

8
magnetization M(r) as the magnetic dipole moment per unit volume, such that the total magnetic dipole
moment of the material is given by Z
m= d3 r M
V
A magnetic dipole experiences a torque in a magnetic field, and this is the origin of both paramagnetism
and diamagnetism. In the case of paramagnetism, it is the dipole formed by the electron spinning about
its axis that experiences the torque, which acts to align the dipole with the field. However, since the Pauli
exclusion principle dictates that electrons in a given atom form pairs with opposing spins, this effectively
neutralizes the torque on the pair, and hence paramagnetism normally occurs in atoms (and molecules)
with an odd number of electrons. In the case of diamagnetism, it is the dipole formed by an electron
orbiting about the nucleus that experiences the torque, which acts to align the dipole antiparallel to the
applied field. Diamagnetism is a universal phenomenon affecting all atoms, but is typically much weaker
than paramagnetism, and is therefore usually observed in atoms with an even number of electrons, where
paramagnetism is normally absent.
The source of BM is therefore the bound current or magnetization current that results from the
combination of all the tiny atomic dipoles or current loops in the material, and we use the symbols
jM and KM for its volume and surface densities, respectively. These can be expressed in terms of the
magnetization as follows:

jM (r) = ∇ × M(r) and KM (r) = M(r) × n̂(r) (r ∈ S)

Similarly, we can view the source of the external applied field that initiates the magnetization process as
free current, such as that flowing through a coil or solenoid, and use the symbol jf for its density. The
total current density that enters Maxwell’s theory is then given by the sum

j(r) = jf (r) + jM (r)

3.2.2 Field Produced by Magnetized Matter


The Biot-Savart formula (in SI units) with jM and KM as sources is

[∇′ × M(r′ )] × (r − r′ ) [M(r′ ) × n̂(r′ )] × (r − r′ )


Z Z
BM (r) = d3 r′ ′
+ dS ′
V |r − r | 3
S |r − r′ |3

where ∇′ indicates that the derivatives are to be evaluated w.r.t. r′ .


Alternatively, we can consider the magnetized matter as a superposition of point dipoles, and deter-
mine the field of these combined. The magnetic field of a point dipole m located at r′ is given by (in SI
units):
1 m · (r − r′ )
 

B(r) = µ0 m δ(r − r ) − ∇
4π |r − r′ |3
or equivalently (whereby evaluation of the gradient gives an extra delta-function term):
 
µ0 8π ′ 3r̂(r̂ · m) − m
B(r) = m δ(r − r ) +
4π 3 |r − r′ |3

If we assign a point dipole moment dm(r′ ) = M(r′ )d3 r′ to every element of the sample volume V ,
then summing their contributions gives

r − r′
Z Z
3 ′ ′ ′ µ0
BM (r) = µ0 d r M(r )δ(r − r ) − ∇ d3 r′ M(r′ ) ·
V 4π V |r − r′ |3
Z Z
µ0 1
= µ0 d3 r′ M(r′ )δ(r − r′ ) − ∇ d3 r′ M(r′ ) · ∇′
V 4π V |r − r′ |

where we note that the first term is only non-zero for r ∈ V (r = r′ ), and in the second line we have used
the fact that
r − r′ 1 1

= −∇ ′
= ∇′
|r − r | 3 |r − r | |r − r′ |

9
If we now define the magnetic scalar potential of a volume distribution of point magnetic dipoles with
density M(r) via Z  
1 3 ′ ′ ′ 1
ψM (r) ≡ d r M(r ) · ∇
4π V |r − r′ |
(N.B. for proof of convergence of electromagnetism integrals of this type, see Electromagnetic Theory,
Stratton, p.171) we can write
(
µ0 M(r′ ) − µ0 ∇ψM (r) r∈V
BM (r) =
−µ0 ∇ψM (r) r∈/ V (since M = 0)

We now define an auxiliary field HM (r) ≡ −∇ψM (r), and since M = 0 outside the magnetized volume,
the above relations can be summarized as the fundamental relation of magnetic matter:

BM (r) = µ0 [HM (r) + M(r)]

Comparing with E = −∇ϕ(r), we can interpret HM (r) as a field produced by a fictitious magnetic
’charge’ with volume density ρ∗ (r) = −∇ · M(r) and surface density σ ∗ (rS ) = M(r)S · n̂(rS ), such that

r − r′ r − rS
Z Z
1 1
HM (r) = d3 r′ ρ∗ (r) ′ 3
+ dS σ ∗ (rS )
4π V |r − r | 4π S |r − rS |3

Note that HM inside V is often called the demagnetization field as it tends to point opposite to M. This
is because the surface charge density σ ∗ = M · n̂ is the source of the field HM .

3.2.3 Example: Uniformly Magnetized Sphere


Consider a sphere with uniform magnetization M = M ẑ, as shown in the Figure below. The volume
densities of both magnetization current jM = ∇ × M and fictitious magnetic charge ρ∗ = −∇ · M are zero
for this system. Hence, we can regard the magnetic field inside and outside the sphere as produced either
by a surface density of fictitious magnetic surface charge σ ∗ = M · n̂ (the plus/minus signs drawn on the
left-hand sphere), or by a surface density of magnetization current KM = M × n̂ (solid lines drawn onto
the right-hand sphere). We shall use σ ∗ to find the magnetostatic potential ψM , and thus the magnetic
field BM , everywhere.

From Laplace’s equation we find that (see Zangwill, p.418):


(
1
Mz r<R
ψM (r, θ) = 31 3 cos θ
3 M R r2 r>R

where θ is the angle that the radius makes with the z-axis. Using HM (r) ≡ −∇ψM (r) we find that
(
− 31 hM r<R
HM (r) = R3 3(r̂·M)r̂−M
i
3 r3 r>R

from which the fundamental relation of magnetic matter yields


(
2
µ0 M r<R
BM (r) = 3
µ0 HM (r) r > R

10
The field lines for both BM (r) and HM (r) are shown in the Figure below. Outside the volume of the
sphere, BM = µ0 HM is identical to the expression for the magnetic field of a point magnetic dipole at
the centre of the sphere:
µ0 3(r̂ · m)r̂ − m
BM (r) =
4π r3
4 3
with dipole moment m = V M = 3 πR M.
Inside the sphere, BM (r) and HM (r) are uniform and anti-parallel, which is because the lines of
BM (r) must form closed loops, while the lines of HM (r) must point away from (or towards) the surface
depending on the sign of the magnetic charge shown in the previous Figure.

3.2.4 Total Magnetic Field


So far, we have only considered the magnetization current density jM = ∇ × M and the magnetic field
BM produced by magnetized matter. No account was taken of other sources of magnetic field, which
were collectively called ’free’ current, with corresponding density jf . We now consider the total magnetic
field B produced by the total current density j = jM + jf . These are the quantities that enter Ampère’s
law, which we write in the form
∇ × B = µ0 j = µ0 (jM + jf ) = µ0 [∇ × M + jf ]
We can then define the auxiliary vector field H(r) (which is the natural generalization of HM when all
sources of magnetic field are taken into account) via
B(r) = µ0 [H(r) + M(r)]
if we make the following identification:
∇ × H = jf
We can also use the above relation to rewrite the other magnetostatic Maxwell equation, ∇ · B = 0, as
∇ · H = −∇ · M = ρ∗
The above two equations are the magnetostatic Maxwell equations in matter, and show that both free
current and fictitious magnetic charge are sources of H(r).
We now illustrate how H(r) can be obtained from the Helmholtz theorem, now that we have expres-
sions for ∇ · H and ∇ × H. These allows us to define the following two functions:
Z ∗ ′ Z ′
1 3 ′ ρ (r ) 1 3 ′ jf (r )
Φ(r) = d r and A(r) = d r
4π V |r − r′ | 4π V |r − r′ |
Starting with the first of these, we begin by rewriting it as
∇ · M(r′ ) M(r′ )
Z Z   Z  
1 1 1 1
Φ(r) = − d3 r′ = − d3 ′
r ∇ ′
· + d3 ′
r M(r′
) · ∇ ′
4π V |r − r′ | 4π V |r − r′ | 4π V |r − r′ |
where in the last step uses Zangwill 1.178. The first term on the RHS vanishes due to the divergence
theorem (see Zangwill 1.182), so that we are left with just the second term, which can be recognized as
just the magnetostatic potential ψM from earlier.

11
Hence, on further application of the Halmholtz theorem we may write H(r) in the form

jf (r′ ) jf (r′ )
Z Z
1 1
H(r) = −∇ψM (r) + ∇× d3 r′ ′
= −∇ψM (r) + d3 r′ ∇ ×
4π V |r − r | 4π V |r − r′ |

Using Ida 8.30 we may rewrite this as

∇ × jf (r′ )
Z   Z
1 3 ′ 1 1
H(r) = −∇ψM (r) + d r ∇ × jf (r′ ) + d3 r ′
4π V |r − r′ | 4π V |r − r′ |

The last term is zero because jf (r′ ) is not a function of the unprimed coordinates, and so the curl is zero.
Using the relation from earlier we may rewrite the second term to yield the following:
′ ′
 
3 ′ jf (r ) × (r − r )
Z Z
1 3 ′ ′ 1 1
H(r) = −∇ψM (r) − d r jf (r ) × ∇ = −∇ψM (r) + d r
4π V |r − r′ | 4π V |r − r′ |3

3.2.5 Simple Magnetic Matter


For most materials, the magnetization M is proportional to the magnetic field, provided the latter is not
too strong. By convention, we express this relation in terms of H rather than B, so that we have

M = χm H

where the (dimensionless) constant χm is known as the magnetic susceptibility of the material, and is
positive for paramagnets and negative for diamagnets. Materials that obey the above relation are known
as linear media. From the definition of H, we obtain the constitutive relation

B = µ0 (H + M) = µ0 (1 + χm )H = µH

where µ is known as the permeability of the material, defined via

µ ≡ µ0 (1 + χm )

In addition, we can define the relative permeability of the material via


µ
µr ≡ 1 + χm =
µ0

12

You might also like