PHAS2201 - Electricity and magnetism

Lecturer: Dr. Mario Campanelli

Disclaimer
Electromagnetism is not a recent subject, and (unfortunately!) this course
does not deal with recent research subjects. The topics discussed here can
(and should!) be found in a large number of books, that will explain the sub-
ject in more depth than these lecture notes, that are therefore NOT meant
as the unique source of information about the course content, but rather as
an invitation to explore more.
Textbooks
I.S. Grant and W.R. Phillips: Electro-magnetism (second edition)
This is the main textbook for this course, covering the whole course
content, as well as having several problems with solutions
D. Griths: Introduction to Electrodynamics
This book is adopted by the follow-up of this course (Electromagnetic
Theories), but the rst half has a large overlap with the content of this
course
R. Feynman: Feynman lectures on physics, part II
Great book to think as a physicist (i.e., out of the box); does not
follow the same order of subjects as the course
R.A. Serway and J.W. Jewett, Jr: Physics for Scientists and Engineers
The big physics book you should already have. Not specically designed
for physicists, but has good worked examples and some standard exer-
cises
1
1 Vectors
1.1 Introduction
Vectors are mathematical objects that represent a set of real numbers in a
multi-dimensional space, that transform according to a given set of rules.
In this course we will only deal with three (or sometimes two-)dimensional
vectors, evolving in an Euclidian space. Diring this course, vectors are indi-
cated by an underlined letter (like A), and their components in a Cartesian
reference frame with the notation (A
x
, A
y
, A
z
), which is a short form for
A
x
x +A
y
y +A
z
z. The magnitude, or length, of a vector is a scalar (a zero-
dimension vector) indicated by |A|; in an Euclidian space, its value is given
by Pythagoras theorem:
|A| =
_
A
2
x
+A
2
y
+A
2
z
.
1.2 Products between two or three vectors
The scalar product of two vectors A and B is indicated by AB = |A||B| cos
where is the angle between the two vectors. So, given their lengths, the
scalar product is maximal when the two vectors are parallel, and zero when
ortogonal. The scalar product is commutative, so A B = B A, and A A =
|A|
2
.
The vector product AB is a vector perpendicular to both, with length
|A||B| sin , and direction going inside the plane dened by A and B in the
clockwise direction. Due to this geometrical asymmetry, the vector product
is anticommutative: AB = AB, so AA = 0 (each vector is parallel
to itself). Geometrically, the length of a vector product represents the area
of the parallelogram dened by the two vectors. In component form, the
vector product can be expressed as a determinant:
A B =

x y z
A
x
A
y
A
z
B
x
B
y
B
z

The sclar triple product is the scalar product between a vector and the vector
product of other two:
A (B C) = B (C A) = C (A B)
2
Geometrically, it represents the volume of the parallelepiped dened by the
three vectors, and in component form it will be the determinant of the matrix
built with the three vectors:
A (B C) =

A
x
A
y
A
z
B
x
B
y
B
z
C
x
C
y
C
z

The vector triple product is less used, and its expansion is

A (B C) = B(A C) C(A B)
1.3 The operator
We can say that an operator behaves like a vector if it requires perform-
ing operations on various dimensions, and behaves like a vector under rota-
tions and translations. In Cartesian coordinates, the dierential operator
(nabla) can be dened in three dimensions as
= (
x
,
y
,
z
) = x
x
+ y
y
+ z
z
It is used in correspondance with vectors and scalar elds to have a short
form for the dierential operratons of gradient, divergence and curl. For a
given scalar eld T(x,y,z), we can dene the gradient as T, the result
being a vector indicating the rate and direction of change of T in the space.
If T is constant, its gradient will be zero, if if it increases along one direction
its gradient will point there. The gradient is a linear operator.
The application of the nabla operator to a vector eld can be done through
the equivalent of a scalar or a vector product. The divergence is the scalar
product (this denition is not mathematically correct, but useful for prac-
tical purposes) of the nabla operator and a vector eld V :
v =
x
v
x
+
y
v
y
+
z
v
z
It can be see as the total rate of change of the vector eld along its direction,
and is connected to sources (emitters or absorvers) of the eld itself.
The curl of a vector eld will be the vector product of nabla and the
eld. Its representation in vector components follows by the denition of
vector product:
v = (
y
v
z

z
v
y
) x + (
z
v
x

x
v
z
) y + (
x
v
y

y
v
x
) z
It is connected to the degree of rotation of the vector eld around each point
in space, and will be parallel (or antiparallel) to the rotation axis.
3
1.4 Combination of operators
The Laplacian is the combination of a divergence and a gradient, so when
applied to a scalar eld the result will be a scalar, while its application to a
vector eld will produce a vector having for each component the laplacian of
the corresponding component of the initial eld:

2
T = = (
2
x
+
2
y
+
2
z
)T

2
v = (
2
v
x
,
2
v
y
,
2
v
z
)
Of the other combined operators, it is important to notice that:
(T) = 0
v = 0
The rst states that the rotation of a vector eld that was obtained from a
gradient of a scalar eld is always zero. This is connected to the notion of
conservative elds. If a vector eld is coming from a gradient, we expect that
its integral between a point A and B will be the dierence of the values of
the initial scalar eld T in those two points, and will be independent on the
integration path. This is only possible for irrotational elds, since otherwise
a line integral from a point A to itself could be dierent from zero.
The second equation states that the curl of a eld cannot have sources.
The curl of a curl is given by the relation
v = ( v)
2
v
1.5 The divergence (Gauss) theorem
We know from the fundamental theorem of calculus that
_
B
A
(
dF(x)
dx
)dx = F(B) F(A)
It can be extended to elds in the space, connecting the line integral of a
vector eld to the values of a scalar eld at the beginning and the end of the
integration path:
_
B
a
(T) dl = T(B) T(A)
4
Notice that this relation involves integrating over a 1-dimensional object to
get a zero-dimensional one. The generalisation to three and two dimensions
(but a similar theorem could be demonstrated in multiple dimensions) is
_
V
( A)dv =
_
S
A ds
where the rst integral is made over the volume of a closed surface, while
the second on the surface that denes the same volume. It states that the
total (outgoing minus ingoing) ux of a vector eld (i.e., integral of the scalar
product of the eld and the normal vector to each innitesimal element of the
surface) is equal to the volume integral inside that surface of the divergence
of the eld. Intuitively, we see that a positive divergence of a eld can be
seen as a source of the eld itself, and therefore crates a positive outgoing
ux of this eld from a surface.
This theorem can be easily demonstrated in the case of a small cube,
aligning the x axis to the direction of the change of the eld, where it reduces
again to the 1-dimensional fundamental theorem. It can be generalised to
the multi-dimensional case using the linearity of the operator, and to the
case of a closed volume of any shape dividing it in an innite number of
innitesimal cubes and pyramids.
1.6 The rotation (Stokes) theorem
Having seen relations between 1- and 0-dimensions quantities (the fundamen-
tal theorem of calculus), one between 2- and 3-dimensions integrals (Gauss
theorem), we discuss now the relation between a line (1-dimensional) inte-
gral and a surface integral. It is known as Stekes theorem, even if all of the
above can acteually be seen as the generalisation o the fundamental theorem
of calculus to multiple dimensions. The classic form of Stokes theorem is:
_
S
A =
_
L
A dl
The surface integral of the curl of a vector eld over all surfaces having
as boundaries a closed line L is equal to the line integrl of the eld along
the line L. This gives us an intuitive interpretation of the curl: if a eld has
nonzero curl, its line integral over a closed line; this is clearly not possible
for a conservative eld.
5
Let us prove this theorem for an innitesimally small square placed on
the xy plane, with sides paralles to the axes (this proof can be seen in Feyn-
mans book). Let us indicate the four sides with 1,2,3 and 4, starting from
the bottom one and moving in anti-clockwise direction; let us indicate as
A
x
(1), A
y
(2), A
x
(3), A
y
(4) the average values of the horizontal and vertical
components of the eld over the four sides. The line integral will therefore
be
_
A dl =
_
A
x
(1)dx +
_
A
y
(2)dy
_
A
x
(3)dx
_
A
y
(4)dy
where the minus signs for the third and fourth sides happen because we are
moving opposite to the directions of the relative axes. In moving from sides
1 to 3 we have to move by a quantity dy along the y axis, so we can say
that A
x
(3) = A
x
(1) +
dAx
dy
dy, following the rst order expansion of Taylors
theorem along the y axis. In a similar way, but considering that from side 2
to 4 we are going backwards in x, we can write that A
y
(4) = A
y
(2)
dAy
dx
dx.
So, the line integral can be written as
_
A
x
(1)dx +
_
A
y
(2)dy
_
A
x
(1)dx
_
dA
x
dy
dydx
_
A
y
(2)dy+
+
_
dA
y
dx
dxdy =
_
(
dA
y
dx

dA
x
dy
)dxdy
The last expression is nothing else than the z component of the curl o A,
integrated over and innitesimal unit of surface. Therefore it proves the
theorem for this very specic case. The proof can be extended to a triangle,
and the sum of an innite number of innitesimal squares and triangles will
prove the theorem for a planar circuit of any shape; some trigonometry is
needed to extend it to the case of non-planar circuits and surfaces.
6
2 Electrostatics
2.1 Coulombs law
Coulombs law states that the force F
21
between two charges q
1
and q
2
at a distance r
21
in is:

F
21
=
1
4
0
q
1
q
2
r
2
21
r
21
(2.1)
where r
21
is the unit vector between the two charges q
1
and q
2
.
The constant
0
is called the permittivity of free space and
is equal to

0
= 8.85 10
12
C
2
N
1
m
2
. (2.2)
7
2.2 The electric eld
The electric eld is a vector eld, where the three components
of a vector depend on the position in each point in space.
Consider once again two charges, q
1
and q
2
, at a distance r
21
.
We have seen that the interaction between the two charges is
described by the Coulomb law, which predicts a force equal to
F
21
=
1
4
0
q
1
q
2
r
2
21
r
21
.
We can say that one of the charges (say q
1
) creates an electric
eld E in space:
E =
1
4
0
q
1
r
2
r . (2.3)
When another charge (in this case q
2
) is introduced, an elec-
tric force acts on it. This force is given by:
F
21
= q
2
E , (2.4)
and we recover Coulombs law. In other words, the electric eld
can be dened as the electric force acting on a charge at
a point in space divided by the magnitude of the charge.
The electric eld has units of newtons per coulomb (N/C).
Consider now the problem of determining the electric eld gen-
erated not by just one charge, but by a group of charges. Since
electromagnetism is a linear theory, the electric eld can be cal-
culated by applying the superposition principle: the total elec-
tric eld due to a group of charges equals the vector sum
of the electric elds of all the charges. Therefore, the total
electric eld E(P) at the point P due to the charges q
1
, q
2
...., q
n
8
is:
E =
1
4
0
n

i=1
q
i
r
2
i
r
i
(2.5)
where r
i
is the distance from the position of the charge q
i
to the
point P.
Consider now the case of a continuous charge distribution. Sup-
pose that within a volume V there is a charge density =
(x, y, z). This means that at the point (x, y, z) there is a charge
of per unit volume.
The charge in an innitesimal volume dV is then dq = dV
and the innitesimal electric eld produced by this charge at
the point R is then
dE(R) =
1
4
0
R r
|R r|
3
(r)dV (2.6)
where r = (x, y, z) is the position of the innitesimal charge
dq = dV . The total electric eld is then obtained by direct
integration over the volume V

E(R) =
_
V
1
4
0
R r
|R r|
3
(r)dV . (2.7)
The treatment above applies also to the case of a charge dis-
tributed over a surface or over a line. In the rst case dq = da,
with the surface charge density, and da the innitesimal
area element. In the second case dq = dl, with the linear
charge density and dl an innitesimal length element.
Exercise: A charge Q is uniformly distributed over a disk of
9
radius R and axis Oz. Determine the electric eld at a point P
on the z axis.
Exercise: Determine the electric eld created by a segment of
length L, carrying a linear charge density , at a point P located
on the medium plane of the wire.
Electric elds can be represented pictorially by electric eld
lines. These lines are parallel to the electric eld vector at any
point in space. The basic properties of these lines are:
The lines must begin on a positive charge and terminate
on a negative charge. If the total charge of the system is
non-zero, some lines will begin or end innitely far away.
E is tangent to the electric eld line at each point. The
direction of the line is the same of that of E;
the number of lines per unit area through a surface perpen-
dicular to the eld lines is proportional to the magnitude
of the electric eld in that region of space.
10
Figure 1: Field lines for a single charge (positive and negative).
Figure 2: Field lines due to two equal positive charges.
11
2.3 Gauss law
Gauss law relates the electric eld on any closed surface to
the net amount of charge enclosed within the surface. We will
see that Gauss law is very useful to determine the electric eld
produced by a charge distribution with a simple geometry.
In order to derive Gauss law we rst introduce two concepts:
the ux of a vector eld, and the solid angle.
1. Flux of a vector eld
The idea of ux of a vector eld is easily explained for a
uid. In this case the vector eld is the velocity v. Consider
a small area a perpendicular to the direction of ow of the
uid (see Fig. 3, left). The uid ux is the rate of ow of
the uid through the area, which is
va . (2.8)
Figure 3: Denition of the ux of a vector eld
If the small area is not perpendicular to v, we have to con-
sider the projection of a onto the plane perpendicular to
the vector v (see Fig. 3), right. Such a projection is equal
12
to a cos . Using vectorial notation, the ux is equal to
v na , (2.9)
where n is the unit vector normal to the surface a. We note
that the sense of the unit vector has to be specied, as n
and n lead to uxes of same magnitude but opposite sign.
For a curved surface, it is necessary to split the surface in
lot of small at surfaces and then sum over these surfaces,
i.e. we have to consider the sum

v n
i
a
i
, (2.10)
which in the limit a 0 becomes

v
=
_
a
v n da . (2.11)
Equation (2.11) generalizes to any vector eld, and in par-
ticular applies to the electric eld E whose ux through a
surface S is

E
=
_
a
E n da . (2.12)
2. Solid angle
In two dimensions, if we have an arc of a circle of radius r
subtended by an angle d, the length of that arc is
dt = rd.
We can use this relation to dene the angle d as
d = dt/r.
13
Note that since dt and r both have dimensions of length,
the angle d is a dimensionless quantity.
In analogous way we can dene a dimensionless solid angle
d in terms of an element of area, da, subtended on the
surface of a sphere

d = da/r
2
(2.13)
14
If we integrate over the surface of the whole sphere we see
that
=
_
sphere
d =
1
r
2
_
sphere
da =
1
r
2
4r
2
= 4 (2.14)
3. Derivation of Gauss law
Now consider a point charge q surrounded by a closed sur-
face, S, of arbitrary shape. From Coulombs law we know
that the electric eld vector E is directed radially outwards
from the charge. Consider an innitesimal area da on this
surface. The unit vector, n normal to this surface will in
general not be in the radial direction. Dene a vector area
da by
15
da = n da (2.15)
ie a vector of magnitude da in the direction of the normal
to the surface.
The electric eld due to the charge q is the usual
E =
q
4
0
r
2
r (2.16)
For an arbitrary shape of surface, E and da will not have
the same direction.
Now consider the total ux of the electric eld through the
surface S given by the integral
_
S
E da =
q
4
0
_
S
r n
da
r
2
(2.17)
=
q
4
0
_
S
cos
r
2
da (2.18)
dacos = da
0
is the area da projected perpendicular to r.
Now the solid angle element d is dened by
d =
da
0
r
2
(2.19)
Hence
_
S
E da =
q
4
0
_
S
d =
q
4
0
4 =
q

0
(2.20)
In the case of many charges contained within the surface,
q
1
, q
2
, . . . q
n
, each gives rise to an electric eld E
1
, E
2
, . . . E
n
.
16
By applying the principle of superposition, we conclude
that the total electric eld, E, at any point is the vector
sum
E =
n

1
E
i
. (2.21)
Then evaluating the same integral as above,
_
S
E da =

i
_
S
E
i
da (2.22)
=

i
q
i
4
0
_
S
r
i
n
da
r
2
i
(2.23)
=

i
q
i
4
0
_
S
d
i
(2.24)
=

i
q
i

0
(2.25)

_
S
E da =
Q
internal

0
, (2.26)
where Q
internal
is the sum of all charges within the surface
S. Equation 2.26 is Gauss law of electrostatics.
Note that the ux through the surface a does not depend
on the shape or size of the surface, only on the amount of
charge it contains.
We have just derived Gauss law for a collection of discrete
charges within a surface S. We wish to extend the law to a
continuous charge distribution. Suppose that within a volume
17
V there is a charge density = (x, y, z). This means that at
the point (x, y, z) there is a charge of per unit volume.
The charge in an innitesimal volume dV is then dV and
the total charge in the volume V is an integral
Q
internal
=
_
V
dV (2.27)
Gauss law for a continuous charge distribution is then

_
S
E da =
_
V

0
dV (2.28)
2.4 Dierential form of Gauss law
Applying the divergence theorem to the electric eld vector E
we nd _
V
( E)dV =
_
S
E ds
Equating the last two equations we get
_
V
( E)dV =
_
V

0
dV
Since this has to be valid for all possible closed volumes (in other
words, no hypothesis has been made on the integration volume),
it is only possible if the integrand is the same, that leads to the
dierential form for Gauss law:
E(x, y, z) =
(x, y, z)

0
18
2.5 Electrostatics in simple geometries
Now let us apply Gauss law to some simple examples of charges
or charge distributions with 1) spherical 2) cylindrical or 3) pla-
nar geometry.
1. Spherical
(a) A uniformly charged solid sphere
Consider a sphere (radius a) with a uniform charge
density . Then the total charge on the sphere is
Q =
4
3
a
3
(2.29)
First, consider a spherical surface S that is concentric
with and encloses the charged sphere, with radius r
a. From the symmetry of the system we can see that
the electric eld is directed radially outwards and is the
same everywhere on the surface S. Let the magnitude
of the electric eld be E.
The LHS of equation 2.26 (Gauss law) is then
19
_
S
E da =
_
S
E da (2.30)
= E
_
S
da (2.31)
= E 4r
2
(2.32)
The RHS of equation 2.26 is just
Q

0
, so that
4r
2
E =
Q

0
(2.33)
therefore

E =
1
4
0
Q
r
2
(r a) (2.34)
which is the same as for a point charge Q.
Second, choose a surface S, inside the sphere, with ra-
dius r < a. Now the RHS of 2.26 is
_
V

0
dV =

0
4
3
r
3
(2.35)
while the LHS is the same as in 2.32 so that
4r
2
E =
4
3
r
3

0
(2.36)
therefore

E =
r
3
0
, r < a (2.37)
20
so the electric eld varies linearly with distance from
the centre of the charged sphere.
(b) A hollow sphere
Consider a thin spherical shell of radius a and negligi-
ble thickness, with a constant charge per unit area, .
The eld outside the shell is the same as for a point
charge or a solid sphere:
E =
1
4
0
Q
r
2
(r a) (2.38)
where Q = 4a
2
is the total charge on the shell.
Inside the shell the eld is zero, as the RHS of 2.26
is zero.
2. Cylindrical geometry
21
Consider an innitely long line of charge (eq a wire with
negligible diameter) with a constant charge per unit length
.
Take a cylindrical surface S, with radius r, length l about
this line.
By symmetry, E must be perpendicular to the line of charge,
directed radially outwards. Also E will be the same at all
points on the curved part of S. Thus, for the top and bot-
tom surfaces of S, the LHS of 2.26 is zero since E n = 0
there. So the LHS of 2.26 is
22
_
S
E da = E 2rl (2.39)
The RHS of 2.26 is
Q

0
=
l

0
(2.40)
so
2rl E =
l

0
(2.41)
and

E =
1
2rl
l

0
=

2
0
r
. (2.42)
Thus E falls of as the inverse of the distance from the wire.
3. Planar geometry
Consider an innite plane of negligible thickness, with con-
stant charge per unit area, .
Take a cylindrical surface S (see Figure) with top and bot-
tom faces of area A.
23
By symmetry, the ux is zero through the curved surfaces
of S, so the LHS of 2.26 is 2 E A. The RHS is
Q

0
=
A

0
.
Therefore

E =

2
0
. (2.43)
So the eld strength is constant and does not depend on
the distance from the (innite) sheet.
A dierent case arises when instead of an innitely thick
surface we consider an insulator, with depth mach larger
than the distance at which we want to measure the eld.
In this case, the ux will only exit the upper surface, and
the LHS of 2.26 is only E A. In this case, due to the
additional charges beneath the surface, the eld will be

E =

0
. (2.44)
24
2.6 Electric potential
Consider the work done in carrying a charge, q
0
, from point
a to point b in an electric eld (see Figure). It is the negative
of the integral of F ds along the path taken

W =
_
b
a
F ds (2.45)
(negative because we do work when we move against the force,
eg if F and ds are in opposite directions, we do work and the
potential energy increases.)
Now the force and the electric eld are related by
F = q
0
E, (2.46)
so that
25

W = q
0
_
b
a
E ds. (2.47)
This is the work done against the electric force to move the
charge q
0
. As energy is conserved, this must equal the change
in potential energy U of the charge-plus-eld system.
Now consider the work done on a unit charge (set q
0
to unity):
W
unit
=
_
b
a
E ds. (2.48)
26
For the case of the eld due to a single charge q, we know
that
E =
q
4
0
1
r
2
r (2.49)
so that
W
unit
=
_
b
a
q
4
0
1
r
2
r ds =
q
4
0
_
b
a
dr
r
2
(2.50)
(since r ds = dr ).
Then

W
unit
=
q
4
0

1
r
a

1
r
b

, (2.51)
which depends only on the endpoints of the movement. We
deduce from the principle of superposition that this is true in
all cases for an electric eld. The line integral does not depend
on the path taken from a to b.
The RHS of 2.51 is the dierence between two numbers. We
can write it as
W
unit
=
_
b
a
E ds = V (b) V (a) (2.52)
where V (x, y, z) is the electric potential at any given point
(x, y, z). Note that equation 2.52 only denes the dierence be-
tween V (a) and V (b), not the absolute value of either.
For convenience, we therefore choose a reference point P
0
(often
taken as innity) where we dene V = 0. We can then write the
potential at a point P as
27

V (P) =
_
P
P
0
=
E ds. (2.53)
This is therefore the work done in bringing a unit charge from
innity to the point P through an electric eld E.
The units of electric potential are therefore joules per coulomb
or volts.
Note that we can choose the path we take from to P
0
, the
result will be the same. In particular if part of the path is per-
pendicular to E then the integral over that part is zero, while if
the path is parallel to E, the contribution is just
_
Eds.
2.7 Electric eld as gradient of the potential
From mechanics we know that the relation between a force F
and the potential energy associated with it, W, is
F = W (2.54)
In the case of an electric eld E, the force on a charge is
F = qE and the electric potential is V = W/q, the work or
energy per unit charge. Therefore
qE = (qV ) (2.55)
therefore

E = V (2.56)
In the previous Sections we introduced the concept of eld
lines, useful to display the electric eld. For the electric poten-
tial, we introduce here the concept of equipotential surfaces,
28
i.e. surfaces characterized by the same potential. We notice that
eld lines always cross equipotential surfaces orthogonally, in
the direction in which the potential decreases most rapidly (since
E = V ).
Remark: the explicit expression for Eq. 2.56 in the various systems of coordi-
nates is:
in cartesian coordinates x, y, z:
E =

V
x
x +
V
y
y +
V
z
z

in cylindrical coordinates , , z:
E =

+
1

+
V
z
u
z

in spherical coordinates r, , :
E =

V
r
u
r
+
1
r
V

+
1
r sin
V

29
2.8 Electric potential for a point charge
For a point charge 2.51 to 2.53 give (as r
a
)

V (x, y, z) =
q
4
0
1
r
(2.57)
where r =
_
x
2
+y
2
+z
2
.
The equipotential surfaces (V = constant) about the point
charge are spheres.
2.9 Electric potential for a discrete charge distribution
For a point charge we know that
V =
q
4
0
1
r
(2.58)
and
E = V =
q
4
0
1
r
2
r (2.59)
where V is the potential at a distance r from the charge q.
For the general case of the potential due to a collection of point
charges, we consider the potential at some point (x
i
, y
i
, z
i
) due
to a set of charges q
j
at (x
j
, y
j
, z
j
). We use the principle of
superposition:

V (x
i
, y
i
, z
i
) =
1
4
0

j
q
j
r
ij
(2.60)
where r
ij
= |r
ij
| = |r
i
r
j
|.
30
2.10 Electric dipole
A dipole is a system with two charges of equal magnitude and
opposite sign separated by a distance 2d.
The potential at point P is
V =
q
4
0

1
r
+

1
r

(2.61)
where, from the cosine rule,
r
2
+
= r
2
+d
2
2dr cos (2.62)
r
2

= r
2
+d
2
2dr cos( ) (2.63)
= r
2
+d
2
+ 2dr cos (2.64)
therefore
1
r

= (r
2

1
2
=
_
r
2

1 +
d
2
r
2

2d cos
r
_

1
2
(2.65)
31

1
r

=
1
r

1 +
d
2
r
2

2d cos
r

1
2
(2.66)
Putting 2.66 into 2.61 gives an exact expression for the po-
tential V , although it is not very easy to dierentiate. A long
way from the dipole, where r d, the terms
x =
d
2
r
2

2d cos
r
(2.67)
are small (1) and we can use the binomial expansion
(1 +x)
n
= 1 +n x +
n(n 1)
2
x
2
+. . . (2.68)
with n =
1
2
to get
(1 +x)

1
2

= 1
1
2
x +. . . (2.69)
so that
1
r
+

1
r

=
1
r
_
1
d
2
2r
2
+
d cos
r
+. . .

1
d
2
2r
2

d cos
r
+. . .
_
(2.70)
therefore
1
r
+

1
r

=
1
r
2d cos
r
(2.71)
and

V
dipole
=
q
4
0
2d cos
r
2
(r d) (2.72)
32
which is the dipole potential at large distances. The quantity
p = 2qd is called the electric dipole moment, and we can
also dene an electric dipole moment vector p, as a vector with
magnitude p, and direction equal to the vector connecting the
negative and the positive charges of the dipole itself. For dis-
tances much larger than d, the potential generated by an electric
dipole at distance r will be
V (r) =
pcos
4
0
r
2
=
p r
4
0
r
3
(2.73)
To be noticed that with respect to the single charge, for which
the potential falls as the inverse of the distance r, for a dipole
the potential falls as 1/r
2
. The electric eld generated by a
dipole, being the derivative of the potential with respect to the
distance, will therefore fall as 1/r
3
as opposed to the case of the
single charge, for which it falls as 1/r
2
.
Exercise (the electric quadrupole): consider the charge cong-
uration formed by a charge 2q at the origin and two charges
33
+q at the points (a, 0, 0). Show that the potential V at a
distance r large compared with a is approximately given by
V = +qa
2
(3 cos
2
1)/4
0
r
3
, where is the angle between
r and the line through the charges.
2.11 Potential for a continuous charge distribution
For a point charge
V =
q
4
0
1
r
; E =
q
4
0
1
r
2
r (2.74)
For the case of a continuous distribution of charge, we apply
the principle of superposition to a small element of charge dq
(see Figure 25)
V =
1
4
0
_
dq
r
(2.75)
34
More explicitly, if the distribution is described by a charge
density (x, y, z), then the potential at (x
i
, y
i
, z
i
) is:

V (x
i
, y
i
, z
i
) =
1
4
0
_
all space containing charge
(x
j
, y
j
, z
j
)
r
ij
d (2.76)
where d = dxdydz is a volume element and r
ij
= |r
ij
| =
|r
i
r
j
|.
Exercise: Find an expression for the electric potential at a point
P located on the perpendicular central axis of a uniformly charged
ring of radius a and total charge Q. Find an expression for the
magnitude of the electric eld at point P.
35
Exercise: A uniformly charged disk has radius a and surface
charge density . Find the electric potential along the perpen-
dicular central axis of the disk. By dierentiating the electric
potential, determine the magnitude of the electric eld along the
same axis.
36
3 Capacitors
Any combination of two conductors carrying charges of equal
magnitude but opposite sign is called a capacitor.
Regardless of the shape of the conductors, they are referred
to as plates. There is a eld E between the plates A and B,
and a potential dierence V = |V
A
V
B
|. Note that V is
always expressed as a positive quantity.
The priciple of superposition implies that the amount of charge
Q on a capacitor (ie on either plate, not the net charge on both
plates, which equals zero) is proportional to V , which we write
as

Q = CV (3.1)
37
where the capacitance
C =
Q
V
(3.2)
is a constant that depends on the shape and separation of the
conductors. C is always positive. It is a measure of a capacitors
ability to store charge and electric potential energy. The SI unit
of capacitance is the farad (F), and one farad = one coulomb
per volt which is a very large unit. In practice values range from
pico (10
12
) to nano (10
9
) to micro (10
6
). One microfarad =
1F = 10
6
F. Note: As the unit of potential dierence V is
the volt, V is often called the voltage eg the voltage between
the plates.
3.1 Parallel-plate capacitors
We have seen in Sec. ?? that for two innite plates, Gauss law
gives
E =

0
(3.3)
where is the surface charge density on either plate.
38
If the plates are of area A and carry a charge q, then =
q
A
,
and the potential dierence
V = Ed =
qd

0
A
(3.4)
where d is the separation of the plates. Note that equation
3.4 corresponds
V = |V (x = 0) V (x = d)| =
_
d
0
Edx = E[x]
d
0
= Ed (3.5)
since E is uniform and in the x-direction.
So, ignoring edge eects (non-uniform E),

C =
q
V
'

0
A
d
(3.6)
which is bigger for larger plates and smaller separations.
39
3.2 Spherical capacitors
Imagine two concentric spherical shells. Note that for r > b,
the net enclosed charge is zero so Gauss law gives E = 0, V
= constant. Between the shells, a r b, the potential is the
same as that of a point charge Q, so that at r = b
V
b
=
1
4
0
Q
b
(3.7)
and the potential of the inner shell (radius a) is
V
a
=
1
4
0
Q
a
(3.8)
Therefore
V = |V
a
V
b
| =
Q
4
0

1
a

1
b

=
Q
4
0
b a
ab
(3.9)
therefore
40

C =
Q
V
= 4
0
ab
b a
(3.10)
We can use 3.10 to derive the capacitance of an isolated spher-
ical conductor: as b ,
b
ba
1, therefore

C
single sphere
= 4
0
a (3.11)
41
3.3 Cylindrical capacitors
A solid conductor of radius a, a cylindrical shell of radius b
and length l. If l b we can neglect end eects and use the
result for an innite line of charge with density , that
E =

2
0
1
r
r a r b (3.12)
where we put = Q/l.
Note the physical assumptions and arguments here -
Neglecting end eects means that the eld is radially out-
ward and perpendicular to the axis.
The charge on the outer cylinder does not contribute to the
eld inside it (Gauss law).
Outside the central conductor, r a, the eld is that due
to a line of charge.
42
The potential dierence between a and b is
V = |V
a
V
b
| =
_
b
a
E dr =
Q
2
0
l
_
b
a
dr
r
=
Q
2
0
l
ln

b
a

(3.13)
therefore
C =
Q
V
=
2
0
l
ln(
b
a
)
(3.14)
Note that the cylindrical geometry is that of a coaxial cable
as used in electricity and signal transmission.
43
3.4 Combinations of capacitors
In electric circuits we often use two or more capacitors either
arranged in parallel or in series, represented as
where the capacitor symbol
reects the common parallel-plate design. Note the dierence
between the capacitor and battery symbols.
Parallel combination
44
What single capacitor has the same capacitance as two others
wired in parallel?
The wires are conductors - they connect things with the same
potential. Thus the potential dierence across the battery ter-
minals V is the same as across the plates of C
1
and C
2
.
V =
Q
1
C
1
=
Q
2
C
2
(3.15)
In the equivalent circuit we want to store the same total
charge
V =
Q
C
parallel
, Q = Q
1
+Q
2
(3.16)
45
Combining these two equations
C
parallel
=
Q
V
=
Q
1
+Q
2
V
=
C
1
V +C
2
V
V
(3.17)
therefore

C
parallel
= C
1
+C
2
(3.18)
The capacitance increases.
Series combination
What single capacitor has the same capacitance as two wired
in series?
In the diagram, the right hand plate of C
2
and the left hand
plate of C
1
are connected to the battery terminals. The battery
46
drives electrons away from the negative terminal so that they ac-
cumulate on the right hand plate of C
2
, giving the charge Q.
An equal and opposite charge must accumulate on the left hand
plate of C
1
.
Now consider the right hand plate of C
1
and the left hand plate
of C
2
. They are connected but isolated from the rest of the cir-
cuit. If no battery is present the net charge on these two plates
is zero and it must remain so once a battery is connected. So
they must carry equal and opposite charges also. The positive
charge on the left hand plate of C
1
creates an electric eld that
attracts negative charge to the right hand plate of C
1
. This pro-
cess will continue until there is electrostatic equilibrium which
occurs when the charges on the plates are equal in magnitude
(but of opposite sign).
The nal conguration has charge of the same magnitude on
all four plates as shown. The potential dierence across each
capacitor may be dierent as shown.
The total potential dierence is
V = V
1
+V
2
=
Q
C
1
+
Q
C
2
(3.19)
will be the same in the equivalent circuit
V =
Q
C
series
(3.20)
therefore

1
C
series
=
1
C
1
+
1
C
2
(3.21)
the capacitance decreases
47
3.5 Energy stored in a charged capacitor?
How much work is done to charge a capacitor?
Imagine that charge is passed from one plate to the other,
against the electric eld between the plates. The work done to
transfer an element of charge dq when the plates carry a charge
q and the potential dierence between the plates is V is
dW = V dq (3.22)
where V = q/C. Remember that V is the work required to
move a unit charge.
The total work done is
W =
_
Q
0
q
C
dq =
1
2
Q
2
C
(3.23)
which equals the total electric potential energy U stored in the
capacitor:

U =
1
2
Q
2
C
(3.24)
48
or, since C = Q/V
U =
1
2
Q V (3.25)
or
U =
1
2
C (V )
2
(3.26)
These results apply to any capacitor, whatever its geometry.
Note that sometimes V is written just as V , so U =
1
2
QV etc.
49
3.6 Energy density
The potential energy stored in a capacitor can be seen as being
stored in the electric eld. eg a parallel plate capacitor, plate
area A, plate separation d. Neglecting edge eects
E =
V
d
(3.27)
The volume of the capacitor is A d, so the energy density, u,
(energy stored per unit volume) is
u =
U
A d
(3.28)
Now
U =
1
2
C (V )
2
=
1
2
d
2
C E
2
(3.29)
We saw that (equation 3.6) we saw that C =

0
A
d
therefore
u =
U
A d
=
1
2
d
2
0
A
d
E
2
a d
=
1
2

0
E
2
(3.30)
This result is much more general than that, and in fact repre-
sents the energy density in vacuum of any electric eld.
50
4 Electricity in matter
We have seen that when a conductor is placed in an external
eld, conduction electrons move until induced charges ensure
that the macroscopic electric eld is zero everywhere inside the
conductor.
In insulating materials, or dielectrics as they are often called,
all electrons are bound to particular atoms, but applied elec-
tric elds still slightly displace the electrons in each atom. This
displacement results in the appearance of induced charges which
reduce the eld in the insulator, though not completely canceling
it. In this chapter we introduce the basic properties of dielectrics
and discuss the microscopic origin of these properties.
4.1 Polarization and dielectric constant
An ideal dielectric material has no free charges. However the
material constituents, atoms and molecules, are aected by the
presence of an electric eld. To be specic, consider for example
a dielectric made of a noble gas like argon. The electric eld
causes the displacement of the electrons and of the nucleus, in
opposite direction. From a macroscopic point of view, this can
be visualized as a displacement of the entire positive charge in
the dielectric relative to the negative charge. The dielectric is
said to be polarized.
To be more quantitative, consider a small element v of a di-
electric medium that is, as a whole, electrically neutral. If the
medium is polarized, than a separation of positive and negative
charge has been eected, and the volume element is charaterized
51
by an electric dipole moment p. The electric polarization,
or simply polarization, of the medium P is dened as the elec-
tric dipole moment per unit volume:
P =
p
v
. (4.1)
4.2 Torque and energy for an electric dipole in an elec-
tric eld
Let us consider a dipole in a uniform eld E:
We dene the electric dipole moment, p as a vector of mag-
nitude qd directed from q to +q.
In a eld the dipole experiences a torque aligning it so that
p is parallel to E.
The torque acting on the dipole is
=

i
r
i
F
i
where the sum is extended to the two charges, and r
i
is the
position of the i-th charge from the center of the dipole.
= r
+
(+qE) +r

(qE) (4.2)
52
= q(r
+
r

) E (4.3)
which can be written

= p E (4.4)
The potential energy U of the dipole is a minimum when p is
parallel to E and a maximum when p is perpendicular to E. The
work done to rotate the dipole from
i
to
f
equals the change
in U:
U
f
U
i
= work done =
_

f

i
||d = pE
_

f

i
sin d (4.5)
U
f
U
i
= p E(cos
f
cos
i
) (4.6)
By convention we choose U
i
= 0 at
i
= 90
o
so that
U
f
= p E cos
f
(4.7)
53
or in vector notation

U = p E (4.8)
4.3 Relations between P and E
The degree of polarization depends on the applied electric eld
E, and on the properties of the constituents (atoms or molecules)
of the dielectric material. We limit here our discussion to materi-
als for which P vanishes when E vanishes. This is the behaviour
of most dielectric materials. We also assume the material to
be isotropic, which implies that the polarization has the same
direction as the electric eld which causes it. Under these as-
sumptions, and in the limit of weak applied electric eld we can
write
P = E (4.9)
where is called the electric susceptibility of the material.
The creation of a polarisation eld made of dipoles will not
produce a net charge inside a homogeneous material, but will
aect the surface boundaries. Lets take for instance a paral-
lelepipedal bloc of a dielectric material, where due to polarisa-
tion all internal dipoles are oriented along the x axis, with the
54
negative charge having smaller x value than the positive one. If
we take a section of this dielectric with dimentions much larger
than the distance between the positive and negative charges of
the dipol, its net charge will be zero; however, the two faces
parallel to the yz axis will have a non-zero surface polarisation,
due to the abundance of negative charges on the face at small x
value, and of positive ones on the face at larger x.
Let us consider now a capacitor lled with a dielectric mate-
rial. Its capacitance in vacuum is given by the ratio of charge
and applied voltage
C
0
=
A
E
0
d
=

0
A
d
where E
0
is the electric eld in vacuum and the charge density
on the plates. Introducing a dielectric, the electric eld will
create a polarisation, and the two faces of the dielectric facing
the capacitors plates will acquire a nonzero surface density. The
electric eld will change, and for a homogeneous dielectric the
new eld will be proportional to the eld in vacuum. Lets call
the proportionality constant, such that
E = E
0
/
and the capacitance is also increased by a factor :
C =
A
Ed
=

0
A
d
(4.10)
Let us now consider a parallelepipedal volume, with two surfaces
parallel and of equal size to the plates of the capacitor, and oth-
erwise very thin containing all the surface charges on one plate
of the capacitor and on the corresponding face of the dielectric.
The charge on the capacitor plate will be Q = A, while the
55
charge due to polarisation will be Q
p
= AP. The total ux
going out of this volume (neglecting the side surfaces) is given
by
A( P)
and because of Gauss theorem
A( P)/
0
= EA
But the charge on the metal plate
= E
0

0
= E
0
so replacing it in the previous expression we have
E = (
0
E P)/
0
and rearranging P and E we get
P = ( 1)
0
E (4.11)
that together with the previous denition of for an homoge-
neous medium, leads to
= + 1 (4.12)
therefore the ratio of the electric eld in vacuum and in a di-
electric, known as the dimensionless dielectric constant , is
just unity increased by the electric susceptibility. If a material
has no polarisation, will be 0, = 1 and we would be back to
the vacuum case.
56
Some values for :
Material
Vacuum 1 exactly
Argon gas 1.00052
Dry air 1.00059 often approximated as 1.0
Silicone oil 2.5
Pyrex glass 5.6
Porcelain 6
Water 80 large!
Strontium titanate 233 large!
Looking at the polarised parallelepipedal block of dielectric
material described above, it is easy to demonstrate that
P =
p
(4.13)
where
p
is the charge density due to polarisation. In general,
polarisation charge has to be considered also when applying
Gauss law to obtain the electric eld:
E = /
0
= (
f
+
p
)/
0
(4.14)
where
f
represents the free charges (those visible from the
outside), as opposed to those arising from polarisation. Since
the latter can be hard to determine, we can dene an additional
eld that will only depend on free charges.
4.4 The Electric displacement eld D
Let us then dene
D =
0
E +P (4.15)
57
where the new vector D is called electric Displacement. In ho-
mogeneous materials, P =
0
E, so
D = (1 +)
0
E =
0
E (4.16)
The relation between elds and charge densities can then be
re-written as:
E = /
0
= (
f
+
p
)/
0
=
1

0
(
f
P) (4.17)

0
E + P =
f
(4.18)
(
0
E +P) = D =
f
(4.19)
So the auxiliary electric displacement eld D has the same
behaviour in matter as the electric eld in vacuum, even without
the annoying factor
0
, and is proportional to E in homogeneous
materials.
The behaviour of the elds E and D diers at the bound-
aries between two materials with dierent dielectric constant
(or one dielectric and the vacuum). At this boundary polari-
sation charges will arise, but in general not free charges. The
easiest case is when the elds are perpendicular to the boundary
surface. A discontinuity will arise for the E and P elds, due
to the building up of polarisation charges, while the eld D is
conserved, since no free charges are present:
58
For the general case, let us consider rst a thin surface across
the boundary between the two materials. Since there are no free
charges, the ux of the D eld going out of this surface has to
be zero, so the components of D perpendicular to the surface
are equal: D
1,
= D
2,
.
59
The eld E will be aected by the presence of the polarisation
charges, but it is easy to see that the component of E parallel
to the boundary between the two materials is conserved. If we
take a small rectangular circuit across the boundary, with the
two sides crossing it innitesimally small, the circuit integral of
the E-eld has to be zero becaus the eld is conserved; therefore
_
E dl = (E
1,k
E
2,k
)l (4.20)
60
and the two components E
1,k
and E
2,k
are equal.
The two elds D and E behave at the boundary of two ma-
terials with dierent dielectric constants like rays of light at the
boundary of two materials with dierent refraction index (it is
no surprise, as we will see later on during this course that re-
fraction index and dielectric constants are connected, and light
is nothing else than a rapidly changing electric and magnetic
eld).
4.5 Electrostatic energy in matter
The inner energy of a capacitor lled with a dielectric is, like in
the vacuum case,
U =
1
2
CV
2
=
1
2
CE
2
d
2
(4.21)
but we have seen that the capacitance is now
C =
0

A
d
(4.22)
so the total internal energy is
U =
1
2

0
AdE
2
=
1
2
ED(Ad) (4.23)
Considering that (Ad) is the volume of the dielectric, and that
this result can be extended to the general case, also when the
two elds are not parallel, we can write the energy density in a
generic point of space
u(x) =
1
2
D(x) E(x) (4.24)
Of course this relation is also valid in vacuum.
61
5 Electrostatic problems
Gauss law in dierential form in vaccum can be written as
E = /
0
(5.1)
We know that the eld and potential are related by the relation
E = (); therefore we can relate potential and charge density
through the Poisson equation:

2
= /
0
(5.2)
which is a second-order dierential equation whose solution will
depend on the location of volume and surface phase densities
(r) =
1
4
0
_
V
(r
0
)d
0
|r r
0
|
+
1
4
0
_
S
(r
0
)d
0
|r r
0
|
(5.3)
where the integral is performed over all space and surfaces; but is
quite complicated, especially when we have to deal with induced
or polarisation charges and non-trivial geometries.
A special case when the electrostatic problem can be easily
solved in the presence of induced charges is when a charge q is
close to an innite at conducting wall. To be noticed that each
conducting surface can be considered as innite and at as
soon as the proba charge is close enough. On a conductor, the
charges will arrange themselves to be at equal potential on the
surface. Since we can always add a constant to the potential
without changing the underlying physics, we can set the value
of the potential at the conducting surface to be zero. This is the
same condition that would be found in the case of a charge -q
placed inside the conducting material in a symmetric position
with respect to the initial charge q.
62
Therefore, without knowing the exact positions of the charges
induced on the conductor, we can calculate the value of the
electric eld and/or of the potential outside the conductor itself
replacing it with this mirror charge inside it:
63
So, eld and potential will not only depend on the distance
of the point under consideration (P in this case) from the main
charge, but also from the mirror charge. Since electromag-
netism is a linear theory, this technique can be applied to several
charges, as long as the approximation that the conducting plane
can be considered as innite and at holds.
64
6 DC circuits
6.1 Electric current
The current through a surface A is dened as the net charge
that passes through the surface per second.
I =
dQ
dt
(6.1)
where dQ is the amount of positive charge that passes
through A.
Recall that in electrostatic equilibrium there are no electric
elds or currents inside a conductor, so no moving charges.
All charges are arranged on the surface. For currents to
exist in the volume of a conductor, there must be a source of
electric eld that drives charges into motion and maintains
that motion.
Question: in the conductor shown in Figure 47, what cur-
rent passes through the planes aa
0
, bb
0
, cc
0
?
Answer: the same current, because of charge conservation.
The SI unit of current is the ampere (A): one ampere =
one coulomb per second (1 A = 1 Cs
1
).
65
6.2 A microscopic model of current
Current ows if a potential dierence (and hence and elec-
tric eld E) exists across a conductor. An electron responds
by moving with a net drift velocity, v
d
in the opposite di-
rection to E, as in Figure 48.
Although it is electrons that move, conventionally we imag-
66
ine equivalent positive charge carriers +q moving with a
velocity v
d
parallel to E.
We dene the current density as a vector in the direction
of v
d
whose magnitude is the net amount of charge crossing
a unit area perpendicular to the drift velocity in unit time,
so

J = nqv
d
(6.2)
with n the number of charge carriers per unit volume.
The current owing through an area A is then

I =
_
A
J da (6.3)
6.3 Ohms law
Georg Ohm (1787-1854) found experimentally that for many
materials, including most metals, the current density is pro-
portional to the applied electric eld,
J = E (6.4)
where is a constant called the conductivity of the conduc-
tor. Equation 6.4 is called Ohms law. The conductivity
, as well as its inverse, the resistivity , dened by
=
1

(6.5)
does not depend on the geometry of the conductor but is a
property of the material from which the conductor is made.
67
Resistivity has units of ohm metres ( m). Some typical
values are:
Material Resistivity, (m)
Copper 1.7 10
8
Aluminium 2.82 10
8
Iron 10 10
8
Carbon 3.5 10
5
(Glass) 10
10
to 10
14
The nal entry shows why glass is used as an electrical
insulator on telephone poles etc.
Ohmic materials are materials that obey Ohms law - they
include the resistors commonly used in electrical circuits.
Non-ohmic materials include semiconductors (see Figure 51),
frequently used in digital logic devices, such as pocket cal-
culators.
68
Consider now the case of a wire of length l and cross-
sectional area A. A uniform eld E is associated with a
constant potential dierence V :
E =
V
l
(6.6)
(this comes from E = V in one dimension, ie E
x
=

V
x
).
Thus we have, combining 6.3, 6.4 and 6.6
J =
I
A
=
V
l
(6.7)

V = I

l
A

= I R (6.8)
69
where
R =
l
A
(6.9)
is the resistance of the conductor. Its SI unit is the ohm
(). One ohm = one volt per ampere (1 = 1 V A
1
).
6.4 Electrical energy and power
Consider the following circuit:
A battery connected by wires (of negligible resistance) to a
resistor of resistance R, maintaining a potential dierence
V across R. From Ohms law a steady current I ows.
The current I corresponds to the charge owing per unit
time, so in a time dt, a charge dQ = Idt ows through he
resistor R.
This charge changes potential by V as it moves through
the resistor and therefore loses electric potential energy
dU = dQV = IdtV (6.10)
70
We dene power as the rate of change of potential energy,
P =
dU
dt
= IV (6.11)
In the circuit, chemical energy from the battery is con-
tinually being converted to internal thermal energy in the
resistor. The course of an electron through R is like that of
a stone falling through water at constant terminal speed.
Its average kinetic energy is constant (current) but it loses
potential energy to its surroundings. In the case of the elec-
trons making up the current, the energy is lost by collisions
with the (static) ions in the metal, generating heat.
The unit of power is the watt (W): one watt = one volt-
ampere = one joule per second (1 W = 1 V A = 1 J s
1
).
6.5 AC and DC current
DC refers to direct current, as opposed to alternating
current (AC).
6.6 Pumping charges: EMF
To maintain a constant current in a circuit we need a charge
pump to keep the potential dierence between a pair of
terminals. Examples of such pumps include batteries and
electrical generators such as dynamos and solar cells.
Consider a simple battery and resistor circuit:
Charges (positive) ow clockwise, from the positive to neg-
ative battery terminals, through R. To pump charge (dq
71
say) from the negative to positive terminals (ie inside the
battery) requires work (dW say).
We dene the EMF (symbol E) of the battery to be the
work done per charge:
E =
dW
dq
(6.12)
where EMF stands for electromotive force the force that
makes the charges move.
Equation 6.12 is essentially the same as the denition of
potential dierence V . Note that E is therefore not a
force but a potential dierence.
6.7 Single-loop circuits
We can apply the principle of conservation of energy to
allow us to calculate currents in a circuit.
72
In section 6.4 we saw that power
P =
dW
dt
= I V (6.13)
In the circuit above V = I R, so
P = I
2
R (6.14)
is the rate of energy dissipation in the resistor. Substituting
in equation 6.13 we have for the energy dissipated in a time
dt,
dW = I
2
Rdt (6.15)
= Edq (6.16)
= EIdt (6.17)
EI = I
2
R (6.18)

E = IR (6.19)
73
The (chemical) energy per charge gained in the battery
equals the (thermal) energy per charge dissipated in the
resistor.
Another way of expressing this is to say that the sum of
all potential dierences around any closed circuit loop must
be zero

closedloop
V = 0 (6.20)
which is known as Kirchhos loop rule.
In the simple circuit composed of a resistor and a battery,
this is simply
E IR = 0 (6.21)
E = IR (6.22)
as in equation 6.19
For resistors in series
Kirchhos loop rule gives
E IR
1
IR
2
= 0 (6.23)
which is like equation 6.19 with R
equivalent
= R
1
+ R
2
. In
general for resistors wired in series
R
equivalent
=

i
R
i
(6.24)
The change in electric potential around such a circuit is
illustrated in Figure 52.
74
6.8 Multi-loop circuits
What is the equivalent resistance for a circuit with resistors in parallel?
We note rst that the potential dierences across R
1
and
R
2
are the same.
We next note that The sum of the currents entering any
junction in a circuit must equal the sum of currents leaving
it

I
in
=

I
out
(6.25)
which is Kirchhos junction rule. It follows from the con-
servation of charge in the circuit.
So in the circuit considered above, and consisting of two
75
resistors in series and a battery, we have:
I = I
1
+I
2
(6.26)
=
V
R
1
+
V
R
2
(6.27)
= V

1
R
1
+
1
R
2

(6.28)
= V
1
R
eq
(6.29)

1
R
eq
=
1
R
1
+
1
R
2
(6.30)
or, in general,

1
R
equivalent
=

i
1
R
i
(6.31)
76
for a collection of resistors wired in parallel.
6.9 Time-varying currents
Figure 53 shows an RC series circuit, a resistor and a
capacitor in series.
To charge the capacitor C from zero, we close the switch S
to point a. From Kirchhos loop rule
E IR
q
C
= 0 (6.32)
where I = I(t) and q = q(t) both vary with time and
across the capacitor V =
q
C
. Note that the potential
decreases on going from the positive to the negative plate
77
of the capacitor. We also have I =
dq
dt
, so equation 6.32
becomes
dq
dt
+

1
RC

q =
E
R
(6.33)
which has the solution
q(t) = CE

1 e
t
RC

(6.34)
from which we can derive the current

I(t) =
dq
dt
=
E
R
e
t
RC
(6.35)
where RC is the time constant of the circuit.
78
Plots of q(t) and I(t) are shown in Figure 54.
Note that at the time t = RC,
I(t = RC)
I(t = 0)
=
E
R
e
1
E
R
=
1
e
= 0.368 (6.36)
The units of RC are F = V A
1
C V
1
= V C
1
s C V
1
= s
79
7 Magnetostatics
7.1 The magnetic eld
Although the phenomenon of magnetism was known about as
early as the 13
th
century BC, and used in compasses it was only
in 1819 than Hans Oersted recognised that magnetism and elec-
tricity are related (current in a wire deects a compass needle).
Nowadays we know that magnetic elds are set up by charges in motion,
as in
1. electromagnets - current owing in a loop or coil of wire,
and
2. permanent magnets - atomic-level current loops (circulat-
ing or spinning electrons) can, in some materials, add to-
gether to give a net magnetism.
In electrostatics we introduced the electric eld E to describe
the electric force F
E
acting on a charge q
F
E
= qE (7.1)
and we saw that the electric eld surrounds any electric charge
q
1
E =
1
4
0
q
1
r
2
r (7.2)
Equation 7.2 applies both to a stationary and a moving charge
but in the latter case the region of space surrounding the mov-
ing charge contains both an electric eld and a magnetic eld
B. This magnetic eld exerts a magnetic force F
B
on any other
charged particle q moving with a velocity v. Experiment shows
that
80

F
B
= qv B (7.3)
where the cross-product shows that F
B
is perpendicular to both
v and B. Note that F
B
never has a component parallel to v,
so it cannot change the speed |v| or the kinetic energy
1
2
mv
2
of
the charge. However it can change the direction of travel of the
charges.
The SI unit for B is the tesla. One tesla = one newton per
(coulomb-metre per second) (1 T = 1 N C
1
m
1
s = 1 N A
1
m
1
)
We can represent B with eld lines, where
the tangent to a eld line gives the direction of B at that
point, and
the spacing of the eld lines represents the magnitude of
B.
81
Figure 55 shows the eld lines near a bar magnet. Note that
they exit one end (called the north pole) and enter at the other
end (the south pole).
When two magnets are brought together, like magnetic poles
repel each other, while opposite poles attract.
Earth has its own magnetic eld due to circulating molten metal
in its core. A compass needle turns with its north pole toward
the Arctic. Although we call it the Earths (geographic) north
pole, it is in fact the Earths magnetic south pole.
Typical eld strengths:
Earths eld 5 10
4
T
Near a small bar magnet 10
2
T
Laboratory superconducting magnet 10 T
7.2 Motion of a charged particle in a magnetic eld
82
1. For a uniform eld, Figure 56 shows a charge moving in
a plane perpendicular to B. Since the magnetic force is
always at right angle to the motion, the force cannot in-
crease the energy of the particle, and its speed |v| will re-
main constant. The particle adopts a circular path with
the magnetic force
F
B
= qv B = qvBr (7.4)
providing the necessary centripetal force
F
centripetal
= ma =
mv
2
r
r (7.5)
to keep it there. Equating 7.4 and 7.5,
mv
2
r
= qvB (7.6)
so the radius of the circle is

r =
mv
qB
(7.7)
and the period T (circumference/speed) is
T =
2r
v
=
2
v
mv
qB
=
2m
qB
(7.8)
The angular frequency (the cyclotron frequency) of the
motion is
= 2f = 2
1
T
=
qB
m
(7.9)
independent of the speed of the particle.
83
2. If the particle moves with velocity v at some arbitrary angle
to the uniform B, its path is a helix (see Figure 57). W can
resolve the velocity of the particle into components parallel
and perpendicular to the eld
v = v
k
+v

(7.10)
The component v

leads to circular motion at the cyclotron

frequency. The component v
k
leads to uniform motion par-
allel to B. The combined motion is helical.
3. For a non-uniform eld, the motion can be more complex.
Figure 58 shows the Earths eld, which channels charged
particles (mostly from the solar wind) to the magnetic poles.
Here, the particles sometimes collide with atoms in the at-
mosphere and cause them to emit light, the Aurora Borealis
and Aurora Australis.
84
7.3 The Lorentz force
A charge q moving with velocity v in both an electric eld E and
a magnetic eld B experiences both an electric force qE and a
magnetic force qv B. The total force is therefore

F = qE +qv B (7.11)
and is called the Lorentz force.
Look at three applications involving crossed elds (E perpen-
dicular to B).
1. The velocity selector
See Figure 59. Only those particles for which the downward
force qE is matched by the upward force qvB will pass
through the selector. Their speed is

v =
E
B
(7.12)
which can be selected by changing E or B.
85
2. The mass spectrometer
In the instrument shown in Figure 60 a second uniform eld
B
0
is placed beyond the end of a velocity selector. From
equation 7.7, the radius of the semi-circular path in Figure
60 is
r =
mv
qB
0
(7.13)
so that
m
q
=
rB
0
v
=
rB
0
B
E
(7.14)
A variation of this method was used by J.J. Thomson in
1897 to measure
m
e
e
for electrons.
86
3. The Hall eect
In 1879, Edwin Hall saw that when a current-carrying con-
ductor is placed in a magnetic eld B, an electric eld E
H
is
generated that is perpendicular to both B and v
d
(the drift
velocity of the charge carriers and the direction of current
ow), see Figure 61.
It arises because the magnetic force pushes the charges to
the edges of the conductor, leading to a Hall voltage, V
H
and hence and electric eld E
H
=
V
H
d
where d is the width
of the conductor.
We measure the sign of the charge carriers from the sign
of V
H
and the drift speed from the magnitude of E
H
. In
equilibrium the magnetic force on the charge carriers and
the electric force due to the Hall eld must balance
qv
d
B = qE
H
(7.15)
therefore
87
v
d
=
E
H
B
(7.16)
Once the drift velocity has been determined and knowing
the dimensions of the conductor, the Hall voltage can be
used to measure magnetic eld strength.
7.4 Magnetic force on a current-carrying conductor
Consider a straight segment of current-carrying wire (length
L, cross-sectional area A, current I - see Figure 62) in a eld B.
The magnetic force on each charge q moving with drift velocity
v
d
is qv
d
B. The total force on the segment is:
F
B
= nqALv
d
B (7.17)
where n is the number of charges per unit volume and there-
fore nqAL is the total charge in the volume AL. Now, in terms
88
of these quantities, the current is
I = nqv
d
A (7.18)
If we dene a vector L as having magnitude L and the direc-
tion of v
d
(ie the direction of the current) we can write
F
B
= IL B (7.19)
Now consider an arbitrarily shaped segment of wire.
From equation 7.19 the force on a small segment of vector
length ds is
dF
B
= Ids B (7.20)
and the total force on the wire is

F
B
= I
_
b
a
(ds B) (7.21)
where a and b are the endpoints of the wire.
Note that if B is uniform we can take it out of the integral
F
B
= I
_
b
a
ds

B (7.22)
89
Furthermore, if the wire is a closed loop,
_
b
a
ds = 0 (7.23)
so F
B
= 0. Thus the net magnetic force on a closed
current loop in a uniform eld is zero.
90
7.5 Torque on a rectangular current loop; magnetic
dipole
Much of the worlds work is done by electric motors. At the
heart of a simple motor (Figure 63) is a rectangular current-
carrying loop in a uniform magnetic eld B. What is the net
torque on the loop?
If the long side of the loop has length a, and the short side
length b, then looking end-on at Figure 63 we see
For the sides of length a, the current is perpendicular to the
magnetic eld so the force has magnitude
|F
1
| = |F
2
| = IaB (7.24)
We dene a unit vector n normal to the plane of the loop and
with direction dened by a right-hand screw rule from the di-
rection of the current. If the angle is the angle between n and
91
B, the torque due to the forces F
1
and F
2
is
= F
1
b
2
sin +F
2
b
2
sin (7.25)
= IabBsin (7.26)
= IABsin (7.27)
where A = ab is the area of the loop.
Note that the forces on the short sides are equal in magnitude
and opposite in direction.
A convenient form of equation 7.27 is obtained by dening the
vector area A = A n when

= IA B (7.28)
Equation 7.28 allows us also to calculate the potential energy
U of a current loop in an external magnetic eld. The torque
tends to decrease the angle , therefore the potential energy
increases with . This xes the sign between the derivative of
92
the potential and the torque:

= = IABsin . (7.29)
By integrating we obtain
U = IABcos + constant . (7.30)
If we choose the constant so that the potential energy U is zero
when is /2 we have
U = IABcos = IA B . (7.31)
In analogy with the electric dipole moment, we dene the mag-
netic dipole moment m of a small coil of area A carrying a
current I to be

m = IA (7.32)
where A is related to the current sens by the right-hand screw
rule. In terms of the magnetic dipole moment, the potential
energy can be rewritten as

U = m B (7.33)
which is the magnetic analogous of the relation U = p E for
the electric dipole.
7.6 Analogies and dierences between electric and mag-
netic dipoles
In magnetism there are no free charges, but only magnetic dipoles,
therefore we can use an analogy with the relations derived for
electric moments. We have seen that for distances much larger
than the distance between the charges of an electric dipole, the
93
electric potential falls as 1/r
2
and the electric eld falls as 1/r
3
,
and both are proportional to the value of the electric moment.
For analogy, we can write that the magnetic eld at a distance
r from a magnetic moment m will be proportional to m and
inversely proportional to r
3
; we can therefore introduce the pro-
portionality constant
0
, permeability of free space, and write
that for a point r on the axis of a magnetic loop with momentum
m,
B =

0
4
2m
r
3
(7.34)
There is however a dierence between electric and magnetic
dipoles, since the former is actually composed of two charges
and at some point the large distance approximation will break
down, while this will not happen in the magnetic case, and this
dierence can be used to derive the basic laws of magnetism.
Let us consider a at surface of electric dipole (for instance, a
parallel plate capacitor). The electric eld lines will be strong
and oriented from the positive to the negative plate in the space
between the two plates, while much weaker (and therefore often
neglected) and going in the opposite direction (always from the
positive to the negative plate), but this time outside the capac-
itor volume. At the surface of the plates there is a discontinuity
of the electric eld, due to the surface charges on the plates
themselves acting as charge sources.
94
Let us consider the potential on the x axis, perpendicular to
the plates of this capacitor and passing through its centre. For
distances much larger than the separation between the plates,
we know the potential can be approximated by
(r) =
Np
4
0
_
S
ds r
r
3
(7.35)
The potential tends to zero both for x +inf and x inf,
so the integral of the electric eld on the x axis will be zero:
_
+inf
inf
E dl = 0 (7.36)
Due to the discontinuity of the electric eld, the potential as
a function of the x position along the axis will have an abrupt
change in correspondence to the plates:
95
Now, we can take the integral of the electric eld on the
closed path dened by the segment (-r,0) to (+r,0) and the half-
circumference of radius r back to point (-r,0):
96
For r going to innity, we have already seen that the integral
over the x axis
_
+inf
inf
E dl = 0. For the half-circumference part,
we know that the eld goes like 1/r
3
, while the length of the
path itself is linear with r, so again this contribution will be
zero. Overall, we can write, as expected from the fact that the
E-eld is conservative, that
_
E dl = 0 (7.37)
over the closed integration path.
Thing are dierent in the case of the magnetic eld. We have
already seen that the magnetic moment was dened as m = IA
where A is the integral over the surface of the small circuit of
the innitesimal surface element ds. We have already seen also
97
the expression of the potential of the electric eld

E
(r) =
Np
4
0
_
S
ds r
r
3
(7.38)
and from the analogy between electric and magnetic dipoles, the
integral of the magnetic eld (notice that it cannot be dened
a potential, since B is not conservative!) is

M
(r) =
_
r
inf
B dl =
I
0
4
_
S
ds r
r
3
(7.39)
So far, the electrical and magnetic cases have a striking simili-
tude. However, the fact that no magnetic monopoles exist means
that nowhere in the integral of the magnetic eld there is a sharp
discontinuity, nor a change of direction of the eld itself:
The integral of B along the x axis will not be 0, but the value
98
obtained integrating the above expression:
_
+inf
inf
B dl =
0
I (7.40)
and since also in this case the continuation to the semi-circumferece
at innity gives a null contribution,
_
B dl =
0
I (7.41)
This result is only valid because the integration path has
been carried through a magnetic dipole, so crossing a circuit
with current; for any path not including wires with current, the
line integral of the magnetic eld would be zero. This brings us
to
7.7 Amperes law
The integral of the B-eld across a closed loop is equal to the
constant
0
times the current owing through the area enclosed
by that path. For zones of space not including moving charges,
the magnetic eld is conservative, since
_
B dl = 0, but this is
not true in the general case, therefore it is not possible to dene
a scalar potential for the magnetic eld. We will see later on
how a dierent kind of potential can be used.
From Stokes theorem, an line integral of a vector over a
circuit can be expressed as the integral of the curl of this vector
over any surface that is dened by that circuit. The left side of
Amperes law becomes then
_
B dl =
_
S
(B) ds (7.42)
99
We can also recall that the current I owing through a surface
is equal to the integral over that surface of the current density
j:
I =
_
S
j ds (7.43)
therefore Amperes law can also be written as
_
S
(B) ds =
0
_
S
j ds (7.44)
that implies
B =
0
j (7.45)
(known as the dierential form of Amperes law) since no as-
sumption has been made on the integration surface S.
7.8 Magnetic vector potential
We have already seen that, being the eld B non-conservative
in the presence of currents, it is not possible to dene a scalar
potential. However the absence of magnetic monopoles means
that the divergence of the magnetic eld is always equal to zero:
B = 0 (7.46)
From the denition of the dierential operators, the divergence
of a curl is also always dierent from zero, so the B-eld can be
dened as the curl of another eld, A:
B = A (7.47)
This equation being the equivalent of the denition of the scalar
potential for the electric eld, A is referred to as the vector
potential of the magnetic eld.
100
As the scalar potential, A is not uniquely dened. If another
irrotational eld A
0
(i.e., such that A
0
= 0 is added to A,
the physical eld B will stay the same (this is analogous, but
more general, to the fact that the electric eld does not change
if a constant is added to the electric potential). An additional
condition has to be imposed to uniquely determine A, and for
steady currents the usual choice is
A = 0 (7.48)
7.9 The Biot-Savart law
The dierential form of Amperes law can be integrated to give
an expression of the vector potential rising from the ow of
current in a region of space. The result is an expression very
similar to the scalar potential of the electric eld for a given
distribution of charges in space:
A(r) =

0
4
_
V
j(r
0
)d
0
|r r
0
|
(7.49)
Using Stokes theorem the integral over the volume can be trans-
formed into an integral over the surface surrounding the moving
charges
A(r) =

0
I
4
_
S
dl
0
|r r
0
|
(7.50)
where I is the curent owing in the wire, and dl
0
is an innites-
imal vector in the direction of the current in the wire ar the
position vector r
0
. Finally, the magnetic eld B is obtained tak-
ing the curl of A
B(r) =

0
I
4
(
_
S
dl
0
|r r
0
|
) (7.51)
101
and the result is the Biot-Savart law:
B(r) =

0
I
4
_
dl
0
(r r
0
)
|ulr r
0
|
3
(7.52)
7.10 Magnetic eld due to a current in a long straight
wire
Consider a length element ds a distance r from a point P.
The cross product
ds r = |ds||r| sin z (7.53)
= dx sin z (7.54)
which on substitution in equation 7.52 gives
dB =

o
I
4
dx sin
r
2
z (7.55)
To integrate this we need to relate the variables r, , x. By ge-
ometry,
sin =
a
r
r =
a
sin
(7.56)
102
and
tan =
a
x
x =
a
tan
(7.57)
(the minus sign is needed since ds is at negative x).
dx =
ad
sin
2

(7.58)
and
dB =

0
I
4
ad
sin
2

sin
sin
2

a
2
(7.59)
=

o
I
4a
sin d (7.60)
If the end points of the wire are at angles
1
and
2
then the total eld is of magnitude
B =

0
I
4a
_

2

1
sin d (7.61)
=

0
I
4a
(cos
1
cos
2
) (7.62)
If the wire is very long so that
1
= 0 and
2
= , then cos
1

cos
2
= 2 and
103

B
v. long wire
=

0
I
2a
(7.63)
The magnetic eld lines (giving the direction of B) are circles
concentric with the wire which lie in planes perpendicular to the
wire. The curled-straight right-hand rule (Figure 64) can help
us remember.
7.11 Magnetic eld due to a circular current loop
See Figure 65. Let us calculate B at an axial point P a distance
x from the centre of a loop of radius R.
Note that every ds on the loop is perpendicular to r. There-
fore
|ds r| = |ds||r| sin

2
= ds (7.64)
Also, note that r
2
= x
2
+R
2
. So, in the Biot-Savart law equation,
the magnitude
104
dB =

0
I
4
|ds r|
r
2
=

0
I
4
ds
(x
2
+R
2
)
(7.65)
The vector dB has components dB
x
and dB
y
, but in sum-
ming over the loop the contributions to dB
y
cancel out. Thus
B = B
x
x, and
B
x
=
_
dB
x
=
_
dBcos =

0
I
4
_
ds cos
(x
2
+R
2
)
(7.66)
But , x, R are constants for all elements on the loop, so we
can take them outside the integral, giving
B
x
=

0
I
4
cos
(x
2
+R
2
)
_
ds (7.67)
But
cos =
R

x
2
+R
2
and
_
ds = 2R (7.68)
therefore
105

B
x
=

0
IR
2
2(x
2
+R
2
)
3
2
(7.69)
Let us now examine two extreme cases: the eld at the center
of the loop, and the eld at a large distance from the loop.
1. Field at the center of the loop.
At the centre of the loop, x = 0, so

B
0
=

0
I
2R
. (7.70)
2. Field at a large distance from the loop
In the denominator of Eq. 7.69 we can neglect R as x >>
R. We obtain in this way
B
x
'

0
IR
2
2x
3
. (7.71)
This can be rewritten in terms of the magnetic dipole mo-
ment m = IA n = IR
2
n of the loop as

B =

0
4
2m
x
3
. (7.72)
7.12 Magnetic force between two parallel conductors
Two current carrying wires will exert magnetic forces on each
other (see Figure 66). Wire 2 creates a eld B
2
at wire 1, which
exerts a force (see equation 7.19)
F
1
= I
1
l B
2
(7.73)
As l is perpendicular to B
2
, |F
1
| = F
1
= I
1
lB
2
.
106
If the wires are very long, then from equation 7.63, B
2
=

0
I
2
2a
and

F
1
=

0
I
1
I
2
l
2a
(7.74)
The direction of the force depends on the currents. If I
1
is
parallel to I
2
the wires attract each other, while if I
1
and I
2
are
anti-parallel, they repel each other.
Note that the force between two parallel wires is used to de-
ne the ampere.
7.13 Amp`eres law
As we have seen, in magnetostatics (steady currents) we can nd
the the net magnetic eld B due to any distribution of currents
107
via the Biot-Savart law
dB =

0
I
4
ds r
r
2
(7.75)
If the distribution is complicated we may need to use a com-
puter to work out the total eld. However, we can write equation
7.75 in a form which, in cases of high degrees of symmetry, is
easily applied. This form is called Amp`eres law:

_
B ds =
0
I (7.76)
where I is the total current passing through any surface bounded
by the closed path around which the line integral
_
B ds is cal-
culated.
We will see how this is applied to a wire, a solenoid and a toroid.
7.14 Magnetic eld due to a current in a long straight
wire
In section 6.7 we used the Biot-Savart law to nd the eld due
to a long straight wire. After a tricky integration we found that
if the wire has a radius R and carries a current I
0
, then the eld
at a distance r from the axis of the wire is of magnitude
B(r) =

0
I
0
2r
(r R) . (7.77)
We can reach this result much more easily from Amp`eres
law, since by symmetry (Figure 67)
_
B ds = B
_
ds = B.2r =
0
I
0
(7.78)
B =

0
I
0
2r
(r R) (7.79)
108
We can also use Amp`eres law inside the wire - path 2 in
Figure 67 - where the current, I, enclosed is
I
I
0
=
r
2
R
2
(7.80)
I =
r
2
R
2
I
0
(7.81)
Then applying Amp`eres law
_
B ds = B.2r =
r
2
R
2

0
I
0
(7.82)

B(r) =

0
I
0
2R
2
r (r < R) (7.83)
so that the eld magnitude varies as shown in the following
gure:
109
7.15 Magnetic eld due to a solenoid
A tightly wound helical coil of wire is called a solenoid (see Fig-
ure 68). The magnetic eld inside a solenoid is nearly uniform
(Figure 69). Outside a solenoid the eld is weak as the eld con-
tributions cancel each other out. For an ideal solenoid (Figure
70), B = 0 outside and B is uniform inside. Amp`eres law then
gives, for the path shown in Figure 70,
110
_
B ds =
_
b
a
B ds +
_
c
b
B ds +
_
d
c
B ds +
_
a
d
B ds (7.84)
Now, along bc and da, B is perpendicular to ds, so the inte-
grals are zero. Along cd, B = 0 and along ab, B is parallel to
ds, so overall we have
_
B ds =
_
b
a
B ds = Bh (7.85)
Then applying Amp`eres law
Bh =
0
I
enclosed
=
0
NI (7.86)
where N is the number of turns in the solenoid. Thus we
have for B,

B =
0
nI (7.87)
where n =
N
h
is the number of turns per unit length.
111
7.16 Magnetic eld due to a toroid
A toroid is like a solenoid bent into the shape of a hollow dough-
nut (Figure 71). By symmetry, the B eld lines are concentric
circles inside the toroid. Consider one of these as an Amp`erian
path of radius r, then
_
B ds = B.2r =
0
NI (7.88)

B =

0
NI
2r
(7.89)
Note that for an ideal toroid, B = 0 anywhere outside the
toroid - including in the hole of the doughnut.
7.17 The magnetic eld due to an innite conducting
sheet
One more example of using Amp`eres law,
_
B ds =
0
I (7.90)
to calculate B. Consider a thin, innitely large sheet in the
yz plane (Figure 73), carrying a current of density J
s
per unit
112
length in the +y direction.
By symmetry we expect B to be directed along z - think of it
as a lot of long wires side by side (see Figure).
Applying Amp`eres law to the rectangular path abcd we get
_
B ds =
_
b
a
B ds +
_
c
b
B ds +
_
d
c
B ds +
_
a
d
B ds (7.91)
But
_
b
a
B ds =
_
d
c
B ds = 0 because B is perpendicular to ds
(7.92)
and
113
_
c
b
B ds =
_
a
d
B ds = Bl (7.93)

_
b
a
B ds = 2Bl =
0
I
enclosed
=
0
J
s
l (7.94)

B =

0
J
s
2
(7.95)
which is independent of the distance from the sheet ie it is
uniform. Note the similarity with the uniform electric eld due
to an innite sheet of charge, E =

2
0
.
8 Magnetic materials
In an analogous way to the polrisation of a dielectric from an
external electric eld, a magnetic eld would generate a mag-
netisation inside a material:
M = N < m > (8.1)
114
For materials with no memory of magnetisation (well see
later what that means), the magnetisation will be directly pro-
portional to the external eld
M =
B
B/
0
(8.2)
In analogy with the electric case, we can dene the magnetic
intensity
H =
1

0
B M (8.3)
As the magnetisation M is the magnetic equivalent of the polar-
isation P, the intensity H is the equivalent of the displacement
eld D.
8.1 Diamagnetism
Let us consider now a semi-classical model of an atom, where
electrons are turning in a circular orbit of radius r with velocity
v. The current due to a single electron is
I =
ev
2r
(8.4)
creating a magnetic moment
m =
evr
2
2r
(8.5)
The angular momentum of the electron will be
L = m
e
r
2
(8.6)
where m
e
is the electron mass; so, accounting for the fact that
the electron charge is negative, the magnetic moment caused
115
by a single orbiting electron will be proportional to its angular
momentum, and equal to
m =
e
2m
e
L (8.7)
and for many electron the magnetic moment will be the sum of
the contributions from each of them.
Without external elds, the Coulombian attraction from the
nucleus will exactly balance the centrifugal force:
m
e

0
r =
Ze
2
4
0
r
2
(8.8)
and the equilibrium frequency is

0
=

Ze
2
4
0
m
e
r
3
(8.9)
If a B-eld parallel to the electron angular momentum is switched
on, the equilibrium of the forces is modied by the presence of
the Lorentz force:
m
e
r =
Ze
2
4
0
r
2
+erB (8.10)
Since the Lorenz force is much smaller than the Coulomb con-
tribution, the orbiting frequency is modied by a small amount,
and using the approximation

1 +x ' (1 +x/2) we nd
'

Ze
2
4
0
m
e
r
3
+
eB
2m
e
(8.11)
The electron gets an extra angular momentumL = m
e
r
2
that creates an extra magnetic moment
< m >=
e
2m
e
m
e
mr
2
=
e
2
r
2
m
e
B (8.12)
116
This result has been obtained assuming that the B-eld is always
parallel to L; averaging over all directions
M =
Ne
2
Zr
2
6m
e
B (8.13)
so if there are no additional eects, the magnetisation will have
an opposite sign to the external eld, and will decrease it inside
the material.
8.2 Paramagnetism
Some atoms or molecules can have their own magnetic moment,
that will be a multiple of the Bohrs magneton
B
=
e~
2m
e
. This
will lead to an intrinsic magnetisation per unit volume m
B
. With
an external magnetic eld, the potential
U
P
= m
B
B (8.14)
is minimised. If the material is at a temperature above abso-
lute zero, the magnetic moments will not all be aligned to the
minimal energy level, due to thermal agitation. The proper
statistical-thermodynamical treatment leads to a magnetisation
M =
Nm
2
B
3KT
B (8.15)
so this eect is positive, and increases the magnitude of the eld
inside the material. Usually, this eect is larger than that of the
change in the angular orbital velocity described in the previous
chapter, so if the material has intrinsic magnetisation, in general
the magnetic eld will be larger than the vacuum case.
117
8.3 Ferromagnetism
An extreme case of paramagnetism is the so-called ferromag-
netism. Some materials like iron exhibit quantum properties
already at room temperature; their electrons, being fermions,
have to obey Paulis exclusion principle, and cannot all occupy
the lowest energy level. For temperatures below a given Curie
temperature (which for metals is of the order of 10000 degrees),
the electrons will behave as they were at the absolute zero, so
according to thermodynamics, all their spins will be aligned in
the same direction. In practice, however, the material will be
divided into macroscopic magnetic domains, each of them with
all electron spins oriented in the same direction, and a stronger
external eld will orient more domains in the same direction.
This phenomenon creates very large magnetisations, with coef-
cients << 1. The relation between the external eld and
the induced magnetisation is however far from linear, apart for
the smallest values of the magnetisation, and since it involves
re-orienting in a permanent way the spins of the various do-
mains, the magnetisation depends on the history of the external
eld, which is quite a unique phenomenon. Let us start with a
metallic material with all domains oriented at random. In this
case, the average magnetisation will be zero, and the B-eld in
absence of an external magnetic intensity it will be zero. In the
plot of B vs H, B(0) = 0.
118
If we apply a current on a circuit close to the material, a
magnetic intensity H will be generated (H only depends on free
currents, so it is independent on magnetisation, it does not de-
pend on ). The B-eld inside the material will increase roughly
linearly with the external magnetic intensity (path from point
0 to point a). Apart from the large value of , this process is
not very dierent from what would happen in a paramagnetic
material. When the H-eld is reduced to zero, however, most of
the domains would still remain in the same position, so the value
of B inside the material would remain dierent from zero even
without external eld (path from a to b). This phenomenon is
phenomenon is known as residual magnetisation. If we want to
remove this residul magnetisation, a reverse H-eld has to be
applied (path from b to c). Furthermore, applying a further
reverse eld, the material will be magnetised in the opposite di-
119
rection, up to a saturation value opposite to that of point a. The
cycle (called hysteresis) can be repeated ad libitum, producing
and removing magnetisation in the material.
8.4 Electromagnets
Since ferromagnetic material have very large values of , they
are used to produce very strong magnetic elds out of circuits
traversed by currents. If we have a solenoid with a coil density
N, from Amperes law
_
H dl = NI (8.16)
H =
NI
2r
(8.17)
For a homogeneous material, B =
0
H so the presence of
the ferromagnetic material will increse the B-eld inside the
solenoid by a factor .
The magnetic eld can be increased even more if a small
opening is made in the metal. If the solenoid is bent in a circle,
with radius r, and a slit size of l, the previous integral becomes
(2r l)
B

0
+
lB

0
= NI (8.18)
120
If the slit is small (l << 2r)
B '

0
NI
2r +l
(8.19)
But also for very large values of , l >> 2r, so
B '

0
NI
l
(8.20)
The eld in the slit can be made even stronger than the eld in
the ferromagnetic material itself.
121
9 Electromagnetic induction
9.1 Magnetic ux
We dene the magnetic ux,
B
, of the magnetic eld B through
a surface S as the scalar given by

B
=
_
S
B da (9.1)
The SI unit for magnetic ux is the weber, where one weber
= one tesla metre squared.
Example: The magnetic ux through a plane of area A placed in a uniform eld
B is

B
= BAcos (9.2)
where is the angle between B and the normal to the plane surface.
Example: Figure 74 shows a rectangular loop beside a current carrying wire. What
is
B
through the planar surface bounded by the loop?
We know that
B =

0
I
2r
(9.3)
122
and that B is directed perpendicular to and into the plane of the loop. Therefore:

B
=
_
S
B da =
_

0
I
2r
da (9.4)
The area element da = bdr (see Figure 74) so

B
=

0
Ib
2
_
a+c
c
dr
r
=

0
Ib
2
log

a + c
c

=

0
Ib
2
log(1 +
a
c
) (9.5)
9.2 Gauss law for magnetism
Our aim is now to determine divB. We start by calculating
div(dB) where, as we saw in the previous chapter, dB is given
by
dB =

o
I
4
ds r
r
3
. (9.6)
In calculating divB, ds is a constant vector. We apply the
general relation
(P Q) = Q (P) P (Q) . (9.7)
As ds is constant, we have

ds r
r
3

= ds

r
r
3

(9.8)
Therefore we are left with calculating (r/r
3
). By writing
the term (r/r
3
) as
r/r
3
=

1
r

we deduce
(r/r
3
) =
1
r
= 0
123
as f = 0 for any f.
We therefore conclude that div dB = 0. If this is true for
every contribution dB, it must also be true for the vector B
itself. We can therefore write

divB = 0 . (9.9)
This is Gauss law for magnetism.
We notice that Eq. 9.9 implies that

B
=
_
S
B da =
_
V
divB dV = 0 (9.10)
for any closed surface S. This means that for a given closed
surface S around a volume V as many lines enter the volume V
as leave it. In other words, there are no magnetic charges
or magnetic poles on which magnetic eld lines end (or start
from), and magnetic eld lines are continuos.
124
9.3 Amp`ere-Maxwell law for time-varying elds
So far we have considered the magnetostatic situation, with
steady currents and E and B unchanged with time. But what if
E, for example varies with time? Consider a charging capacitor
(Figure 75). A conduction current I ows to the positive plate,
but there is no current between the plates. Consider two surfaces
S
1
and S
2
, both bounded by the path P. Amp`eres law states
_
B ds =
0
I (9.11)
where I is the current owing through any surface bounded by
P. For S
1
, the RHS of 9.11 is
0
I while for S
2
the RHS is
zero. Maxwell concluded that, in the presence of time-varying
electric elds, Amp`eres law is incomplete. Maxwell postulated
a displacement current, dened as
I
d
=
0
d
E
dt
(9.12)
where
E
is the electric ux and revised Amp`eres law to
125

_
B ds =
0
I +
0

0
d
E
dt
(9.13)
This is known as the Amp`ere-Maxwell law.
For the capacitor in Figure 75, assuming a uniform eld E =
Q

0
A
between the plates, the electric ux through S
2
is

E
= EA =
Q

0
(9.14)
and the displacement current is then
I
d
=
0
d
E
dt
=
dQ
dt
(9.15)
which is precisely the same as the conduction current I =
dQ
dt
through S
1
. Thus with the Amp`ere-Maxwell law, the same result
is obtained with S
2
as with S
1
. Note that the major new concept
in equation 9.13 is that magnetic elds B can be produced
both by conduction currents and by time-varying elec-
tric elds.
9.4 Faradays law of electromagnetic induction
We now turn our attention to electric elds produced by chang-
ing magnetic elds.
The experiments of Faraday, carried out in 1831, showed that
a transient current occurred in a closed circuit when
the magnetic ux through the circuit was changed. The
change in magnetic ux can be produced in various way. In the
rstexperiment, a pulse of current is seen in a loop of wire when
a magnet is moved toward or away from it.
In the second experiment, a current pulse is induced in the
secondary coil when the switch in the primary circuit is opened
126
or closed.
Faradays conclusions were:
when the magnetic ux through a circuit is changing, an
electromotive force is induced in the circuit;
127
the magnitude of this e.m.f. is proportional to the rate of
change of the ux.
These results are summarized by the Faradays law of induction

E =
d
B
dt
(9.16)
where

B
=
_
S
B da (9.17)
is the magnetic ux through the circuit in which the EMF, E is
induced.
Example: EMF in a conducting bar on rails Consider
the following Figure: a circuit comprising a conducting bar mov-
ing along two parallel conducting rails with constant velocity v
(eg the axle of a train). In the presence of a uniform B into
the plane, what EMF is induced in the bar? The magnetic ux
through the circuit is

B
=
_
S
B da = B.l.x
Therefore using Faradays law,
E =
d
B
dt
= B.l.
dx
dt
= Blv (9.18)
If the resistance of the bar and rails are small compared to the
overall resistance R of the circuit, we nd the induced current
in the circuit to be
I =
|E|
R
=
Blv
R
(9.19)
128
Example: EMF in a rotating bar
Consider a conducting bar rotating with angular speed =
v
r
in a uniform magnetic eld B perpendicular to the plane of
rotation.
A segment of the bar, of length dr, generates an EMF, which
from 9.18 is
dE = B.v.dr = B..r.dr (9.20)
so the total EMF across the whole bar is

E = B
_
l
0
rdr =
1
2
Bl
2
(9.21)
129
9.5 Lenzs law
Soon after Faraday proposed his law, Heinrich Lenz devised a
rule for nding the direction of the induced current in a loop.
An induced current has a direction such that the magnetic
eld due to the current opposes the change in the magnetic ux
that induces the current
As the EMF is in the same direction as the induced current,
this corresponds to the minus sign in Faradays law.
9.6 Self-inductance
Faradays law tells us that any circuit that carries a time-varying
current (and associated time-varying magnetic ux
B
) has in-
duced in it an EMF E
L
that opposes the source EMF. This eect
is called self-induction, and E
L
is called the self-induced or back
EMF
E
L
=
d
B
dt
(9.22)
If we think now about immobile circuits (ie no moving parts),
130
in which
d
B
dt

dB
dt

dI
dt
(9.23)
where I is the source current, we can write 9.22 as
E
L
= L
dI
dt
(9.24)
where L is a constant, known as the inductance of the circuit.
The archetypal source of induction (or inductor) in a circuit
is the ideal solenoid. Lets consider a circuit containing one
when S is closed, current starts to ow. At any given instant
the eld in the solenoid is
B =

0
NI
l
(9.25)
where
N
l
is the number of turns per unit length. The magnetic
ux through each turn is

each turn
B
= B.A =

0
NAI
l
(9.26)
131
where A is the area of the loop. The total ux through the
solenoid is

total
B
= N.
each turn
B
=

0
N
2
AI
l
(9.27)
and hence the induced EMF is
E
L
=
d
B
dt
=

0
N
2
A
l
dI
dt
(9.28)
so, by comparison with 9.24

L =

0
N
2
A
l
(9.29)
which depends only on the geometry of the solenoid.
The SI unit of inductance is the henry. One henry (H) = one
volt-second per ampere: 1 H = 1 VsA
1
.
9.7 RL circuits
An inductor in a circuit opposes changes in the current through
that circuit. Consider the RL circuit: a resistor and an induc-
tor in series.
When S is closed the back EMF E
L
is induced in the inductor.
Applying Kirchhos loop rule to the circuit
E IR L
dI
dt
= 0 (9.30)
which is a dierential equation in I(t), whose solution is
I =
E
R
(1 e

) (9.31)
132
where is the time-constant of the RL circuit, given by
=
L
R
. (9.32)
From the graph of 9.31 we nd that I
E
R
as t , and
that at t = , I = (1
1
e
).
E
R
= 0.63
E
R
. We notice that with no
inductor in the circuit the current would have gone to
E
R
almost
instantaneously.
133
9.8 Energy stored in a magnetic eld
If we multiply 9.30 by I and rearrange it we can obtain
EI = I
2
R +LI
dI
dt
(9.33)
All the terms in this equation have units of power or energy
per unit time. Thus we can read the equation as
The rate of energy supply by the battery, EI
equals the rate of energy dissipation in the resistor, I
2
R
plus the rate of energy storage in the inductor, LI
dI
dt
.
We can use this last term to dene the energy U stored in
the inductor at any time:
dU
dt
= LI
dI
dt
(9.34)
Therefore
U =
_
U
0
dU =
_
I
0
L.IdI (9.35)

U =
1
2
LI
2
(9.36)
Note the similarity of form to the energy stored in the elec-
tric eld of a capacitor, U =
1
2
Q
2
C
.
In the case of an inductor which is an ideal solenoid, we have
L =

0
N
2
A
l
, and the magnetic eld occupies a volume A.l,
134
so that we can write an expression for the energy density
of the eld
u =
U
A.l
=
1
2

0
N
2
A
l
I
2
A.l
=
1
2

0
N
2
I
2
l
2
(9.37)
But we know that, for a solenoid, B =

0
NI
l
, so we can
write

u =
B
2
2
0
(9.38)
This result is generally applicable. Note again the similarity
of form compared to the energy density of an electric eld
u =
1
2

0
E
2
.
9.9 Mutual inductance
135
If two coils are placed next to each other, as in Figure 86,
the time-varying current in one produces a time-varying
B
in the other, and hence an induced EMF.
It can be shown that
E
1
= M
dI
2
dt
, E
2
= M
dI
1
dt
(9.39)
where M is the mutual unductance.
9.10 Oscillations in an LC circuit
We have three circuit elements: resistance, R, capacitance,
C, inductance, L. We have previously considered the RC
and RL series circuits, now we look at the LC circuit:
Suppose that the capacitor is fully charged to a charge of
Q
max
, and that at time t = 0, the switch S is closed. So
long as there is no resistance in the circuit, oscillations are
set up, see Figure 87.
136
At any time t the energy stored in the electric eld of the
capacitor is
U
E
=
Q
2
2C
(9.40)
and the energy stored in the magnetic eld of the inductor
is
U
B
=
1
2
LI
2
(9.41)
The total energy of the circuit, U = U
E
+ U
B
, remains
137
constant with time, so
dU
dt
= 0. So
dU
dt
=
d
dt

Q
2
2C
+
1
2
LI
2

(9.42)
=
Q
C
dQ
dt
+LI
dI
dt
(9.43)
= 0 (9.44)
But,
I =
dI
dt
and
dI
dt
=
d
2
Q
dt
2
(9.45)

Q
C
dQ
dt
+L
dQ
dt
d
2
Q
dt
2
= 0 (9.46)

d
2
Q
dt
2
=
Q
LC
(9.47)
Note that this dierential equation has the same form as
that governing simple harmonic motion. If we dene
=
1

LC
(9.48)
equation 9.47 becomes
d
2
Q
dt
2
=
2
Q (9.49)
which has solutions
Q = Acos(t +) (9.50)
where is a phase angle and A is a constant.
In our circuit, the charge is at a maximum, Q
max
at t = 0,
so A = Q
max
, and the solution is simply
Q = Q
max
cos(t) (9.51)
138
and the current is therefore
I =
dQ
dt
= Q
max
sin(t) (9.52)
or
I = I
max
sin(t) (9.53)
Graphically, the variations of Q and I are
We can verify that the total energy in the circuit is constant,
since
U =
Q
2
2C
+
1
2
LI
2
(9.54)
=
Q
2
max
2C
cos
2
(t) +
1
2
LI
2
max
sin
2
(t) (9.55)
But since
I
2
max
=
2
Q
2
max
=
Q
2
max
LC
(9.56)
so
U =
Q
2
max
2C
(cos
2
t + sin
2
t) =
Q
2
max
2C
(9.57)
139
9.11 Damped oscillations in an RLC circuit
Adding resistance to an LC circuit is like adding friction to
a simple harmonic oscillator, energy is lost and the oscilla-
tions are damped down. (see Figure 88).
In such an RLC series circuit:
the rate of energy loss as heat in the resistor is
dU
dt
= I
2
R (9.58)
which must be equal to the rate of change of energy in the
other elements of the circuit. We saw above that this has
the value
dU
dt
=
Q
C
dQ
dt
+LI
dI
dt
(9.59)
140
Equating 9.58 and 9.59 we have
Q
C
dQ
dt
+LI
dI
dt
= I
2
R (9.60)
or
LI
dI
dt
+I
2
R +
Q
C
dQ
dt
= 0 (9.61)
or, in terms of Q,
L
dQ
dt
d
2
Q
dt
2
+

dQ
dt

2
R +
Q
C
dQ
dt
= 0 (9.62)
and, dividing through by
dQ
dt
,
L
d
2
Q
dt
2
+R
dQ
dt
+
Q
C
= 0 (9.63)
The solutions to this equation depend on the relative sizes
of R, L and C. If R is small, the circuit is lightly damped
and the solution is
Q = Q
max
e

Rt
2L
cos(
d
t), (9.64)
141
where

d
=

1
LC

R
2L

2
(9.65)
.
as in Figure 88. If R exceeds a critical value R
c
, given by
R
c
=
_
4L
C
(9.66)
the circuit is said to be overdamped and no oscillations
occur.
Graphically,
142
9.12 Dierential form of Faradays law
Consider _
E ds (9.67)
for the path ABCDA shown.
The electric eld at A has components E
x
, E
y
, E
z
. Since
the path lies in the xz plane, there will be no contribution
to E ds from E
y
(since E
y
is x, z). Along AB and CD,
only E
x
will contribute. Along BC and DA only E
z
will
contribute. The eld E
x
at the four corners is
A E
x
B E
x
+
_
E
x
x
_
dx
C E
x
+
_
E
x
x
_
dx +
_
E
x
z
_
dz
D E
x
+
_
E
x
z
_
dz
From this we can see that along CD, E
x
exceeds the value
along AB by
_
E
x
z
_
dz at all points, so the net contribution
143
to
_
E ds from AB and CD is

E
x
z

dz. dx
where the minus sign arises because the path is in the neg-
ative x-direction along CD.
Similarly, the contribution from BC and DA will be

E
z
x

dx. dz
So that overall we have

E
z
x

E
x
z

dxdz
This is the y-component of curl E = E.
E =

i

j

k

z
E
x
E
y
E
z

(9.68)
The magnitude of the magnetic ux through the loop ABCD
is
B
y
dxdz
So Faradays law gives for y-components
(E)
y
=
B
y
t
and similarly for the other components, to get

E =
B
t
(9.69)
144
which is Faradays law in dierential form. Note that we
can change from a total dierential with respect to time
to a partial dierential since the change in ux arises only
through a change in B.
9.13 Dierential form of the Amp`ere-Maxwell law
By following the same procedure outline above, it is possible to
recast the Amp`ere-Maxwell law
_
B ds =
0
I +
0

0
d
E
dt
=
0
_
J da +
0

0
d
dt
_
E da (9.70)
in dierential form

B =
0
J +
0

0
E
t
(9.71)
9.14 Amp`ere-Maxwell law in matter
In matter, magnetisation currents j
M
induce a magnetisation
eld through
j
M
= M (9.72)
and polarisation charges will produce currents
j
P
=
P
t
(9.73)
We can re-write the Amp`ere-Maxwell equation splitting the cur-
rent density into the various terms: free, magnetisation and po-
larisation:
1

0
B = j
f
+j
M
+j
p
+
0
E
t
(9.74)
145
Using the above relations for j
M
and j
p
, we can write
1

0
B = j
f
+M +
P
t
+
0
E
t
(9.75)
so,
(
B

0
M) = j
f
+

t
(P +
0
E) (9.76)
that leads to the generic formulation of the Amp`ere-Maxwell
law in matter:
H = j
f
+
D
t
(9.77)
146
10 AC Circuits
10.1 Why use alternating current
The oscillations in an RLC series circuit can be prevented
from damping out if an external EMF
E = E
max
sin t (10.1)
is used to replace the energy dissipated as heat in the re-
sistor.
Circuits in homes, oces and factories (including countless
RLC circuits) receive such an EMF from electricity gener-
ating companies: in the UK, E
max
= 230 V and the period
of oscillation is T =
2

=
1
50 Hz
=
1
50
seconds.
This EMF delivers an alternating current (AC):
I = I
max
sin(t ) (10.2)
where is a phase angle.
What does this look like at the microscopic level?
The drift speed of electrons in household wiring is approx-
imately 4 10
5
ms
1
. If we reverse their direction every
1
100
sec, the electrons can move only about 4 10
7
m in a
half-cycle, which is about 100 atomic spacings before their
motion is reversed. Is this still useful?
Yes, because if we consider a plane in the conductor a large
current can still ow across it if there are enough electrons
in motion. Energy can still be dissipated in a resistance,
even if the electrons continually reverse their direction of
motion.
147
AC has other useful features:
It is easy to generate
It is easy to transform (change voltage)
It is easy to use - for example to drive motors, where
the oscillation of the current is essential and to deliver
energy, in light bulbs for example.
10.2 AC generators and transformers
Figure 89 shows a simple model for an AC generator. As
the conducting loop is forced to rotate through the eld B,
the sinusoidal EMF of equation 10.1 is generated.
Figure 90 shows an ideal transformer. Two coils with dif-
ferent numbers of turns - N
p
for the primary coil, N
s
for
the secondary coil - are wound around an iron core. The
148
primary coil is connected to an AC source of EMF, symbol
as shown -
which induces an alternating magnetic ux
B
within the
iron. Faradays law gives:
E
p
= V
p
= N
p
d
B
dt
(10.3)
149
and E
s
= V
s
= N
s
d
B
dt
(10.4)
for the induced voltage drop across the secondary, so that

V
s
=
N
s
N
p
V
p
(10.5)
When N
s
> N
p
the output voltage exceeds the input volt-
age (step-up), and when N
s
< N
p
the output voltage is
less than the input (step-down).
10.3 Three simple circuits
We will consider, in turn, three simple circuits comprising
an AC EMF generator and a load of either a resistor, a
capacitor or an inductor. In doing so we will introduce
phasor diagrams to help us in more complex circuits where
we will want to combine sinusoidal currents and voltages
with dierent phases.
10.3.1 A resistive load
See Figure 91. Kirchhos loop rule must be valid for all t,
so
E V
R
= 0 (10.6)
where V
R
is the instantaneous voltage across the resistor.

V
R
= E
max
sin t = V
max
sin t (10.7)
150
since the maximum potential dierence across the resistor
V
R
is the same as that across the source of EMF, E
max
.
The instantaneous current in the resistor is
I
R
=
V
R
R
=
V
max
R
sin t (10.8)

I
R
= I
max
sin t (10.9)
Both V
R
and I
R
vary with sin t (see Figure 91), they are
in phase with each other.
10.3.2 A capacitative load
See Figure 92. Kirchhos loop rule again gives
E V
C
= 0 (10.10)
151
Hence

V
C
= V
max
sin t (10.11)
Now, from the denition of capacitance (C =
Q
V
C
), the
instantaneous charge on the capacitor is
Q = C.V
C
= C.V
max
sin t (10.12)
so the current in the circuit is
I =
dQ
dt
= C.V
max
cos t (10.13)

I = I
max
sin(t +

2
) (10.14)
152
We see that the potential dierence across the capacitor
and the current in the circuit are

2
radians out of phase
(see Figure 92). The current leads the voltage by

2
radians.
The maximum current
I
max
= .C.V
max
(10.15)
can be written in terms of an equivalent resistance, called
the capacitive reactance, X
C
,

I
max
=
V
max
X
C
, X
C
=
1
C
(10.16)
X
C
has the units of ohms.
Note: the phasor diagram is a way to represent the sinu-
soidal variations in current and voltage in terms of rotating
vectors of magnitude I
max
and V
max
.
10.3.3 An inductive load
See Figure 93. Kirchhos rule again gives

V
L
= V
max
sin t (10.17)
From the denition of inductance,
V
L
= L
dI
L
dt
(10.18)
dI
L
=
V
max
L
sin tdt (10.19)
I
L
=
V
max
L
_
sin tdt =
V
max
L
cos t (10.20)
153

I
L
= I
max
sin(t

2
) (10.21)
so V
L
and I
L
are again

2
radians out of phase but now
the current lags the voltage by

2
(see Figure 93).
The maximum current, I
max
=
V
max
L
can be written in
terms of an equivalent resistance, the inductive reactance,
X
L
,

I
max
=
V
max
X
L
, X
L
= L (10.22)
X
L
has units of ohms.
154
10.4 The RLC circuit
We will now consider AC circuits comprising dierent com-
binations of R, L and C in series with an AC source with
voltage given by
E = E
max
cos t . (10.23)
Note that we are using cos t rather than the sin t of the
previous sections. The reason will shortly become apparent.
As it is a series circuit, the same current ows in all parts
of the circuit,
I = I
max
cos(t ) (10.24)
and our aim will be to determine I
max
and the phase angle
in a given circuit.
It is useful to use complex notations in the calculations.
Using the relation
exp(jt) = cos t +j sin t ,
155
where
1
j
2
= 1, we may replace the cosine and sine func-
tions by complex exponentials on the understanding that
we are interested in the real or imaginary parts respectively.
We begin by writing the time-variation of the EMF,
E = E
max
exp(jt) (10.25)
where the bold face, E denotes a complex number and we
understand that the real physical EMF is
E = <(E) (10.26)
The current in the circuit will also vary with angular fre-
quency so we can write
I(t) = I
0
exp(jt) (10.27)
where the presence of the complex I
0
means that the current
and voltage may not vary in phase as we shall see later. The
real physical current will be
I(t) = <(I
0
exp(jt)) . (10.28)
Kirchhos loop rule is now
E RI L
dI
dt

Q
C
= 0 . (10.29)
Dierentiating through with respect to time we have
dE
dt
= R
dI
dt
+L
d
2
I
dt
2
+
I
C
. (10.30)
1
We note that the symbol j rather than i is used in the context of electromagnetism
in order to avoid confusion with current
156
Now using equations 10.25 and 10.27, we have
jE
max
exp(jt) = jRI
0
exp(jt)
2
LI
0
exp(jt)+
I
0
C
exp(jt) .
(10.31)
Dividing through by j we nd
E
max
exp(jt) = I
0
exp(jt)
_
R +j

L
1
C
_
.
(10.32)
This equation has the form
E = IZ (10.33)
where
Z = R +j

L
1
C

= R +j(X
L
X
C
) (10.34)
is the complex impedance of the circuit. The complex impedance
can be written in exponential form as
Z = Z exp i (10.35)
where Z is the magnitude of the complex quantity Z
Z =
_
R
2
+ (X
L
X
C
)
2
(10.36)
and the angle is given by

= tan
1
X
L
X
C
R
(10.37)
The current can be written as
I =
E
max
Z
exp j(t ) (10.38)
157
The voltage leads the current if X
L
> X
C
(peaks in cur-
rent occurr after the peaks in voltage), and lags behind the
current if X
L
< X
C
.
Note that by changing the driving frequency of the EMF
source we can tune the circuit. For example, at large
we can be sure that X
L
= L > X
C
=
1
C
, and that the
voltage leads the current.
At the resonant frequency,
0
, we make X
L
= X
C
so that

0
L =
1

0
C
(10.39)
and hence

0
=
1

LC
(10.40)
At the resonant frequency
the impedance equals the resistance, Z = R,
the voltage is in phase with the current,
the current has its maximum value, I
max
=
V
max
Z
=
V
max
R
10.5 Power in the RLC series circuit
The energy source in the circuit is the AC EMF generator.
This energy is
stored in the electric eld of the capacitor,
stored in the magnetic eld of the inductor, and
158
dissipated as thermal energy in the resistor.
In the steady state, the average energy stored in the ca-
pacitor and the inductor is constant, so the net transfer of
energy is from the source of EMF to the resistor. The rate
of energy transfer, ie the power, is then
P = IV = I
max
cos(t )V
max
cos t (10.41)
This is the instantaneous power delivered by the EMF to
the resistor, as a function of time.
The average power delivered can be found by rst noting
that:
cos(t ) = cos t cos + sin t sin (10.42)
so that equation 10.41 becomes
P = I
max
V
max

cos
2
t cos + sin t cos t sin

(10.43)
and secondly, noting that the time average of sin
2
t is
1
2
,
while the time average of sin t cos t =
1
2
sin(2t) is zero.
The average power is therefore

P
av
=
1
2
I
max
V
max
cos (10.44)
159
10.6 Average power in terms of rms current
Equation 10.44 can be written in terms of the root mean square
(rms) values for current and voltage:
I
rms
=
_
average value of I
2
(10.45)
=
_
average value of I
2
max
sin
2
(t )(10.46)
=
_
1
2
I
2
max
(10.47)

I
rms
=
1

2
I
max
(10.48)
and similarly,

V
rms
=
1

2
V
max
(10.49)
so that

P
av
= I
rms
V
rms
cos (10.50)
160
where cos is called the power factor. Now, in an RLC
circuit:
I
max
= V
max
/|Z| (10.51)
tan
1
= (X
L
X
C
)/R (10.52)
which implies
cos =
I
max
R
V
max
=
I
rms
R
V
rms
(10.53)
which, when substituted in equation 10.50 gives

P
av
= I
2
rms
R (10.54)
which restates our original assumption that, in the ideal
RLC circuit there is no power loss in either the inductor or
the capacitor.
10.7 Resonance in the RLC series circuit
As already discussed, the RLC circuit is said to be in res-
onance when the current is at its maximum value. For the
rms current
I
rms
=
V
rms
Z
(10.55)
this occurs when the impedance
Z =
_
R
2
+ (X
L
X
C
)
2
(10.56)
is a minimum, which is when X
L
= X
C
, corresponding to
=
0
, the resonance frequency where

0
=
1

LC
(10.57)
161
The average power shows a resonance peak since
P
av
= I
2
rms
R =
V
2
rms
R
Z
2
(10.58)
which can be written as

P
av
=
(V
rms
)
2
R
2
R
2

2
+L
2
(
2

2
0
)
2
(10.59)
This expression shows that at resonance, =
0
,
P
av
=
(V
rms
)
2
R
(10.60)
and is at a maximum.
The sharpness of the P
av
() curve is described by the quality factor,
Q,
Q =

0

(10.61)
where is the full width at half-maximum of the curve.
A high-Q circuit responds only to a narrow range of fre-
quencies.
The receiving circuit of a radio is an application of a reso-
nant circuit. Tuning the radio involves varying a capaci-
tor, which alters the circuits resonant frequency to match
that of the incoming electromagnetic wave.
10.8 Power transmission
To get power from a generating station (eg a hydroelec-
tric plant) to household supplies, the EMF is passed along
162
large conducting cables called transmission lines. To min-
imise power losses in the transmission lines ( P = I
2
R),
low resistance metals are used (eg copper), coupled with
relatively small currents (about 100 A). To keep the trans-
mitted power high, high voltages are used (about 440 kV).
163
11 Maxwells Equations
11.1 Summary of Maxwells equations
All electromagnetism can be summarised by the Maxwells equa-
tions and some denitions. The equations for the generic case
(electricity and magnetism in matter) are
_

_
S
D da = Q Gauss
0
law (11.1)
_
S
B da = 0 Gauss
0
law in magnetism (11.2)
_
E ds =
d
B
dt
Faraday
0
s law (11.3)
_
H ds = I
f
+
d
D
dt
Ampere Maxwell law (11.4)
We have also seen their dierential forms:
_

D = Gauss
0
law (11.5)
B = 0 Gauss
0
law in magnetism (11.6)
E =
B
t
Faraday
0
s law (11.7)
H = j
f
+
D
t
Ampere Maxwell law (11.8)
In homogeneous media,
D =
0
E (11.9)
H = B/
0
(11.10)
Finally, a particle with charge q in an electromagnetic eld
is subject to the force
F = q(E +v B) (11.11)
164
11.2 Waves in free space
To summarise, Maxwells equations in dierential form in free
space (no charges or currents, nor polarisation nor magnatisa-
tion present) are
E = 0 (11.12)
B = 0 (11.13)
E =
B
t
(11.14)
B =
0

0
E
t
(11.15)
Now, taking the curl of 11.14 we get
(E) =
(B)
t
=
0

2
E
t
2
(11.16)
where we have used equation 11.15.
We also have a general result for (B) which gives
(E) = ( E) ( ) E (11.17)
But E = 0 in empty space, so
(E) =
2
E (11.18)
and hence

2
E =
0

2
E
t
2
(11.19)

2
is the Laplacian operator

2
x
2
+

2
y
2
+

2
z
2
We can, by the same approach, obtain a similar equation for B
165

2
B =
0

2
B
t
2
(11.20)
Equations 11.19 and 11.20 are three-dimensional wave equa-
tions. In general, the solution of a wave equation has the form
E = E(r vt) (11.21)
where v is the velocity of the wave, and its absolute vaue is
the inverse of the square root of the term multiplying the time
derivative. In this case, electromagnetic radiation moves in free
space at speed
c =
1

0
(11.22)
which is a fundamental constant of nature, namely the speed of
light in vacuum. It is interesting to notice that in a material
with dielectric constant and magnetic constant , the speed
of light is reduced by a factor

(even if after the magnetic
term is neglected since non-ferromagnetic materials have ' 1),
giving rise to the rifraction index responsible of several optical
phenomena.
11.3 Planar waves
The simplest solution to the wave equation is a function equal
to zero along one of the axis, and propagating along another
(planar wave). Each periodical function can be decomposed in
a Fourier sum of sine and cosine terms, so if we take one of them
(in complex representation), we can write the solution for the
electric eld oscillating along the x azis, and propagating along
z:
E = xE
0
exp[i(t kz)] (11.23)
166
with = ck and k = 2/. Using the dierential form of
Faradays law, we can nd the value of the magnetic eld that
accompanies the electric eld:

B
t
= [ xE
0
exp[i(t kz)] = yikE
0
exp[i(t kz)]
(11.24)
B = y
x

E
0
exp[i(t kz)] = y
E
0
c
exp[i(t kz)] (11.25)
In general, a planar wave moving along z will be polarised along
a given direction in the xy plane:
E = ( xE
0x
+ yE
0y
)exp[i(t kz)] (11.26)
and the associated B-eld wil be
B =
1
c
(

k E) (11.27)
11.4 Energy of an electromagnetic wave
We have seen that the energy of an electromagnetic eld is given
by
U =
1
2
_
V
(E D +B H)d (11.28)
In free space, and for the electric and magnetic elds of a wave
travelling along the z axis, this equation becomes
U =
1
2
_
[
0
E
2
0
cos
2
(t kz) +
0
H
2
0
cos
2
(t kz)dxdydz
(11.29)
Integrating over a box of sides a (along x), b (along y) and ct
(along z), and taking the average over a period T
U =
ab
2T
_
T
0
_
ct
0
(
0
E
2
0
cos
2
(tkz)+
0
H
2
0
cos
2
(tkz)]dzdt =
(11.30)
167
ab
2
_
ct
0
(
1
2

0
E
2
0
+
1
2

0
H
2
0
)dz =
abct
4
(
0
E
2
0
+
0
H
2
0
) (11.31)
But H
0
= B
0
/
0
and B
0
= E
0
/c, so H
0
= E
0
/c
0
, and
U =
abc
2

0
E
2
0
=
ab
2
E
0
H
o
(11.32)
so, the energy density crossing a unit surface in unit time is
E
0
H
0
/2, along the direction of movement of the wave, in this
case z. For a general wave, the energy carried by an electro-
magnetic wave per unit time per unit area will be equal to the
Poynting vector
N = E H (11.33)
11.5 DAlembert equation and retarded potentials (non
examinable)
If we write the magnetic eld in terms of the vector potential,
combining Faradays law and the relation between electrostatic
eld and potential, we get a general expression for the electric
eld
E =
A
t
(11.34)
that combined with Gauss law gives an extension of Poisson
equation for non-static elds:

2


t
( A) =

0
(11.35)
To express this equation in a simple form, we recall that adding
an irrotational eld to A does not change the observable B, and
we have the freedoma to x A according to a condition, that for
168
static elds is usually taken as A = 0. For changing elds,
we use instead the so-called Lorentz gauge
A =
0

t
(11.36)
and dening the DAlembert operator
=
2

1
c
2

2
t
2
(11.37)
the generalised Poisson equation becomes the DAlembert equa-
tion for the scalar potential
=

0
(11.38)
Similarly, the combination of the vector potential denition with
the Ampere-Maxwell relation yields

2
A +
0

2
A
t
2
=
0
j (11.39)
So if we dene the four-vectors
A

= (, A) (11.40)
and
j

= (, j/c) (11.41)
the two DAlembert equations for A and can be written in
a compact form that is equivalent to the Maxwell equations,
therefore to basically the whole content of this course:
A

= j

/
0
(11.42)
We remeber that in electrostatic, the solution of the Poisson
equation was
(r) =
1
4
0
_
V
(r
0
)
|r r
0
|
dr
0
(11.43)
169
The solution of the DAlembert equation is very similar, with
the dierence that the potentials do not depend on the present
densities and currents, but on charge densities and currents at
the time when the radition was emitted:
(r, t) =
1
4
0
_
V
(r
0
, t |r r
0
/c|)
|r r
0
|
dr
0
(11.44)
A(r, t) =

0
4
_
V
j(r
0
, t |r r
0
/c|)
|r r
0
|
dr
0
(11.45)
For this reason, once the limited propagation speed of electro-
magnetic interactions is taken into account, the elds will de-
pend on the eects that created them, at the proper times. This
is the reason why we see radiation from the stars not as they
are now, but as they were many millions of years ago.
170