SECTION 6

Magnetic Fields in Matter

This section (based on Chapter 6 of Griffiths) deals mainly with how magnetic effects are modified in the
presence of a magnetic material. The topics are:
• Magnetization
• The field of a magnetized object
• The auxiliary field H
• Linear and nonlinear media

Magnetization
All magnetic field effects are attributed to moving charges (currents). Matter can acquire a magnetization (i.e., a
magnetic dipole moment per unit volume) when atomic dipole moments align. There are three main types of
materials:
Diamagnets acquire a very weak magnetization opposite an external applied magnetic field, and lose their
alignment when the field is removed.
Paramagnets acquire a weak magnetization aligned with an external applied magnetic field, and also lose their
magnetization when the field is removed.
Ferromagnets have dipoles which can align with an external magnetic field to produce a much stronger
magnetization, and also they retain the magnetization after the field is removed.
Paramagnets and diamagnets are simpler, because the individual dipole moments do not interact strongly with
each other. This results in them being linear (the magnetization is proportional to the applied field); we start
with a model of these.
In ferromagnets the individual dipole moments interact strongly (due to quantum mechanics), so they are more
complicated and they are nonlinear; we consider them later.
Torques and forces on magnetic dipoles
It is simplest to look first at a rectangular loop carrying a current and consider other shapes later.

The loop is initially assumed in the xy plane with its centre at the origin. Then it is
z rotated around the x axis so that the normal makes an angle θ with the z axis (and
B M
with B).
If the magnetic field B is uniform, the forces on the two slanted sides are equal and
I
opposite (pulling out on the sides of the loop), BUT
y the magnetic forces on the other two sides create a torque.

!
x
z The  torque  on  the  loop  (tending  to  rotate  it  about  the  x  axis)  is:  
I
     
F where  a  is  the  length  of  the  loop  on  the  slanted  side.  The  force  on  each  segment  
!
is  of  magnitude:    
                                                       
F
where  b  is  the  length  of  the  horizontal  sides  of  the  loop.    So  
I
 
N = IabBsin ! x̂ = mBsin ! x̂ OR N = m!B
Although the expression for N was derived for a rectangular loop, the equation is valid for any current
distribution in a uniform field. (Why? Discuss in class).
In a material the torque tends to line the magnetic dipoles up with the direction of the field (because N = 0 when
m and B are parallel vectors). It is the mechanism responsible for paramagnetism.

  49  
 A necessary condition is (obviously) that m must be nonzero for the atoms or molecules. Hence those with an
  odd number of electrons tend to be paramagnets: the spins of the odd electrons line up with the field.

Force on a magnetic dipole

If the field is uniform, the net force on a current loop is zero, because:
F = I! " (dl ! B) = I !" (dl ! B = 0
( )
However, if the field is nonuniform, there will generally be a net force on the loop.
I
For example, if a current loop is placed above a solenoid,
field B has a radial component because of the fringing
fields, so there is a net downward magnetic force.

B B The magnitude of the net force (vertically down) will be:

F F where θ is the angle of F with the vertical

In general, for an infinitesimal loop with dipole moment m in a field B, it can be shown that the net force is:
F = !(m " B)
Note that this expression looks just like the force on an electric dipole in a nonuniform electric field:
F = !(p " E)
In fact, ideal magnetic and electric dipoles have a very similar behaviour, but the physical origins of the dipoles
are quite different.
+q
I
!q
Real electric dipole Real magnetic dipole
Effect of a magnetic field on an atomic orbit
In a semi-classical model of atoms, the electrons moves around the nucleus in an orbit, which is like a tiny
current loop. In reality, the currents are not simple loops, nor are they steady (since there is only one charge).
In practice, however, the electrons move so quickly in their orbits that we can pretend the current is steady, and
we can approximate the paths as circles. The equivalent current is:
e ev
I = =
T 2! R
where the period T = 2πR / v with radius = R and speed = v.
The magnetic dipole moments is of magnitude m = I πR2 = evR / 2 ,
or vectorially (taking the loop to be in the xy plane):
1
m = ! evR ẑ
2
The dipole moment experiences a torque in a magnetic field, but we have already taken account of this as giving
rise to paramagnetism. Another effect is that the electron speeds up or slows down, depending on the field
direction.
In the absence of a magnetic field, the electric force keeps the electron in its orbit:
e2 v2
= me
4!" 0 R 2 R

  50  
In a magnetic field (which we assume for simplicity to be along z perpendicular to the plane of the orbit), there
is an extra Lorentz force acting on the electron. We now have
e2 v!2
+ ev!B = me
4!" 0 R 2 R
where v’ is the new speed (and we assume R does not change much).
Now we subtract the above two equations to get:
(v!2 " v 2 )
ev!B = me
R
If the speed difference Δv = v’ – v is relatively small, then
(v!2 " v 2 ) = (v! + v)(v! " v) # 2v $v
This gives
eRB
!v =
2me
The dipole moment also changes because of the change in speed:
1 e2 R 2
!m = " e(!v)R ẑ = " B
2 4me
The dipole moment changes in the direction opposite to B, and this turns out to be true whatever the direction of
B relative to the electron loop.
This is the mechanism for diamagnetism. It affects all atoms, but is typically weaker than the paramagnetic
effect, so it is only observed when the atoms have a net zero magnetic moment. Only atoms that have an even
number of electrons tend to be diamagnetic (and only some of them).
In fact, diamagnetism is actually a quantum-mechanical effect, and so the preceding classical calculation is
wrong in detail; still, it illustrates the effect.
The essential property of diamagnetism is that the magnetic dipole moment is in the opposite direction to the
applied B field, whereas for paramagnetism the magnetic dipole moment is in the same direction as the B field.
Magnetization
When the magnetic dipoles in a material align (whether it is with or against the external magnetic field), the
material is said to be magnetized. The magnetization vector M is a measure of the degree of alignment. By
analogy with the definition of polarization P in a dielectric medium, we define
M = Magnetic dipole moment per unit volume
Both paramagnetic and diamagnetic materials experience forces in magnetic fields. In practice these turn out to
be very small compared with the ferromagnetic materials to be discussed later (where M is much larger)..

The field of a magnetized object

The field from a magnetized object
By analogy with the treatment in the electric field case, we look first at the magnetic field produced by a
magnetized object, in the absence of any external field.
We start from the magnetic potential from a single magnetic dipole:

re
       
M To  find  the  potential  due  to  a  magnetized  object,  we  integrate  over  
d!’ little  volume  elements,  each  with  dipole  moment  M  dτ’:                            

     

  51  
This can be used directly to obtain the correct result for the magnetic potential. However, recalling the electrical
case where we replaced the polarized material by bound charges, we will follow a similar method to calculate
A(r). As before (Section 4), we use the fact that:
# 1 & r̂
"!% ( = e2
$ re ' re
with the prime indicating differentiation with respect to re, and then we get
µ * $ 1 '-
A(r) = 0 0 ,M(r!) " #!& )/ d" !
4! + % re (.
Integrating by parts:
µ ., 1 & M(r!) ) .0
A(r) = 0 - $ [ "! # M(r!)] d! ! % $ "! # ( + d! !1
4! /. re ' re * 2.
We can eventually express the second term on the right as a surface integral (it involves some more vector
mathematics and then the use of the divergence theorem). The results is:
µ0 1 µ 1
A(r) = $ ["! # M(r!)] d! ! + 0 !$ [M(r!) # da!]
4! re 4" re
Potential of a volume current Potential of a surface current
Jb = ! " M K b = M ! n̂
With these definitions, we have:
µ 0 J b (r!) µ 0 K b (r!)
4" ! re
A(r) =
4!
" re
d! ! + " da!

So, instead of integrating over the magnetization, we can instead calculate the bound surface and volume
currents and calculate the field directly from them.
This is a direct parallel to the bound surface and volume charges we used in the electrostatic case for the electric
field of a polarized object.
Bound currents
Like bound charges, bound currents are real, and not just a mathematical construct.

Surface bound currents are a result of the little current loops in a magnetized
M object failing to cancel at the boundary: the net effect is like a current flowing
around the boundary of the object.
The surface current must always be normal to both the magnetization of the
material and to the surface, hence (as found before):
I
         
Volume bound currents occur when   the magnetization is nonuniform. Suppose we look at the net current in the x
direction, due to a difference in magnetization in the y and z directions:
z At the interface between the two current loops, the net current in the x direction is:

                     
This corresponds to a current density contribution of:

y
                         
x Similarly, any change in magnetization in the z direction would produce a
corresponding term in the net x current, giving
 

  52  
 !M z !M y
(J b ) x ="
!y !z
The right side is just the x component of a curl. Extending this to 3 dimensions, we get the expected result:
Jb = ! " M

The auxiliary field H

Ampere’s law in magnetized materials
Now that we have the field due to the bound currents, we can move on to get the full field, due to the free
currents and the magnetization of matter. The total current (with bound and free terms) is:
J = Jb + J f
We can write Ampere’s law as:
1
(! " B) = J = J b + J f = (! " M) + J f
µ0
or, collecting the two curl terms:
$1 '
! " & B # M) = J f
% µ0 (
It is convenient to define a new field H for the term in the parentheses, so the result becomes
"1 %
H = $ B ! M'
# µ0 & OR
B = µ 0 (H + M)
Ampere’s law can then be written:
!"H = Jf
In integral form it is:
! H " dl = I f #enc

Therefore we can write Ampere’s law in terms of the free current only. Also, when symmetry allows it, we can
calculate H from Ampere’s law.
Comments on H
H is called the auxiliary field by Griffiths; in many other books it is called the magnetic intensity. Often it is
loosely referred to the magnetic field, but this term is properly reserved for B.
While Ampere’s law in terms of H looks similar to Ampere’s law in terms of B, they can’t be used the same
way, because we need to take account of the divergence as well as the curl. While the divergence of B is always
0, the divergence of H is not necessarily 0, and
! " H = #! " M
Boundary conditions for H
From the previous boundary conditions found for B, we can easily deduce the corresponding boundary
conditions for H as:
! ! ! !
H above " H below = "(M above " M below )
for the component perpendicular to the surface, and
|| ||
H above ! H below = K f " n̂
for the component along the surface (note that it is only affected by the free current in the parallel component).

Linear and nonlinear media

Magnetic susceptibility
For linear magnetic materials (usually paramagnets and diamagnets) , the magnetization is proportional to the
applied magnetic field.
  53  
By analogy with the electric field case (where we used P proportional to E to define an electrical susceptibility),
we might expect that M proportional to B would be used to define a magnetic susceptibility. This is NOT what
happens. Instead, the convention is to use the H field and define
M = !m H
The dimensionless proportionality constant χm is called the magnetic susceptibilty.
For paramagnetic and diamagnetic materials, χm is always much less than one: usually around 10-5.
Magnetic Permeability
For linear magnetic media:
B = µ 0 (1+ ! m )H
This means that the magnetic field B is also proportional to H, so we can write
B=µH
where
µ = µ 0 (1+ ! m )
is called the permeability of the material. Most linear materials have a permeability close to the permeability of
free space.
Even though B and H are proportional inside linear media, we cannot conclude that the divergence of H always
vanishes. To see this, we consider:
0 = ! " B = ! " (µ H) = µ! " H + H " (!µ )
This implies that
1
!"H = # H " (!µ )
µ
This will vanish inside the magnetic material (if µ is constant), but on the boundary it will be very large because
there is a discontinuity of µ.
The volume bound current in a homogenous linear material is proportional to the free current density:
Jb = ! " M = ! " ( ! m H) = ! m! " H = ! m J f
Hence, if no current flows through the material, all bound currents will be at the surface.
Ferromagnetism
Like in a paramagnet, the dipoles in a ferromagnet tend to align parallel to an external magnetic field. However,
a major difference is that in a ferromagnet the magnetic dipoles on one atom interact strongly with the dipoles
on neighbouring atoms (giving an extra tendency to line up). By contrast, in a paramagnet these interactions are
negligible. The extra interactions in ferromagnets are mainly quantum-mechanical and are called exchange
interactions.
Because of these differences, the magnetization M in ferromagnets can be much greater (by several orders of
magnitude) than in paramagnets. Also, unlike paramagnets, the tendency for each dipole to align with its
neighbours means that ferromagnets keep their magnetizations when the external field is removed.

In an unmagnetized piece of iron, the alignments among dipoles occur

in small domains: all the dipoles are aligned in the domain, but each
domain is randomly oriented with respect to the others.
 

When a magnetic field is applied to an unmagnetized piece of ferromagnetic material, the domains which are
magnetized in the direction of the field grow at the expense of their neighbours magnetized in the opposite
direction.
When the B field is removed, the domains do not shift back all the way to their original state (i.e., the process is
irreversible).
  54  
 B B

The magnetization of a ferromagnet (such as iron) depends not only on the applied magnetic field at that instant,
but also on the history. If we place a piece of iron in a coil, and apply a current, the magnetization traces out a
hysteresis loop as shown above.
Notes: The magnetic field from the magnetized ferromagnetic material is much greater than the magnetic field
applied. This is because typically
! m >> 1
If the temperature is increased, random thermal motions tend to disorder the magnetic alignment. For
ferromagnetic materials, there is a particular temperature (called the Curie point), above which the material
becomes paramagnetic instead of ferromagnetic (e.g., for iron, this is 770ºC).

Magnetostatic Problems: Summary for sections 5—6

Usually, we have a current distribution and want to find the magnetic field (sometimes we find the potential
first).
System Method for solving
• Many line currents   • Biot-Savart Law and superposition  
• Continuous current distribution with • Ampere’s Law  
cylindrical, plane, or solenoid symmetry    
• Extended distribution with other • Integrate over current distribution to get
symmetry   vector potential  
• Want potential far from current • Multipole expansion (vector potential of
distribution   dipole)  
• Magnetized material   • Find bound currents and calculate field  
• Linear material in magnetic field   • Find field H from Ampere’s law, calculate
B.  

  55