110A Formulas Solutions

Electricity and Magnetism: Physics 110A
Written by Roger L. Grifth @UCB fall 2007

January 23, 2008
Chapter 1
Vector Analysis
1.1 Vector Operations
The Dot product of two vectors is dened as
AB ABsin n
where is the angle they form when placed tail-to-tail. Note that AB is itself a scalar (hence the
alternative name scalar product). The dot product is commutative,
A B = B A
and distributive,
A (B+C) = A B+A C
if two vectors are parallel, then A B =AB. In particular, for any vector A,
A A =A
2
if A and B are perpendicular, then AB =0.
The cross product is dened as
AB = ABsin n,
where n is a unit vector pointing perpendicular to the plane of Aand B. The cross product is distributive,
A(B+C) = (AB) +(AC)
but not commutative. In fact,
(BA) =(AB)
geometrically, |AB| is the area of the parallelogram generated by A and B. If two vectors are parallel,
their cross product is zero. In particular,
AA = 0
1
1.2 Vector Algebra: Triple Products
(i) Scalar triple product: A (BC). Geometrically, |A (BC)| is the volume of the parallelepiped
generated by A, B, and C, since |BC| is the area of the base, and Acosis the altitude. Evidently,
A (BC) = B (CA) = C (AB)
for they all correspond to the same gure. Not that alphabetical order is preserved. The nonalpha-
betical triple products,
A (CB) = B (AC) = C (BA)
have the opposite sign. In component form,
A (BC) =
A
x
A
y
A
z
B
x
B
y
B
z
C
x
C
y
C
z
(ii) Vector Triple Product: A(BC). The vector triple product can be simplied by the so called
BAC-CAB rule,
A(BC) = B(A C) C(A B)
1.3 Position, Displacement, and seperation vectors
The location of a point in three dimensions can be described by listing it Cartesian coordinates (x, y, z) .
The vector to that point from the origin is called the position vector:
r x x+y y+z z
its magnitude is given as
r =
x
2
+y
2
+z
2
is the distance to the origin, and
r =
r
r
=
x x+y y+z z
x
2
+y
2
+z
2
is a unit vector pointing radially outward. The innitesimal displacement vector, is given as
dl = dx x+dy y+dz z
the seperation vector from a source point to the eld is given as
r r
its magnitude is
=
r r
and a unit vector in the direction from r
to r is
=
=
r r
r r
2
in Cartesian coordinates,
= (x x
) x+(y y
) y+(z z
) z
=
(x x
)
2
+(y y
)
2
+(z z
)
2
=
(x x
) x+(y y
) y+(z z
) z
(x x
)
2
+(y y
)
2
+(z z
)
2
1.4 Problems with Solutions
Problem # 1
The Equation giving a family of ellipsoids is
u =
x
2
a
2
+
y
2
b
2
+
z
2
c
2
Find the unit vector normal to each point of the surface of these ellipsoids.
Given a three-dimensional surface we can nd what the normal vector is by using
u =
u
u
and knowing that
u =
u
x
x +
u
y
y +
u
z
z
=
2x
a
2
x +
2y
b
2
y +
2z
c
2
z
and for the magnitude of this vector we nd
u = 2
x
2
a
4
+
y
2
b
4
+
z
2
c
4
thus the unit vector u is given as
u =
x
a
2
x +
y
b
2
y +
z
c
2
z
x
2
a
4
+
y
2
b
4
+
z
2
c
4
Problem # 2
For each of the following vector elds A (1) determine whether A is constant, (2) nd the divergence
and the curl A, and (3) nd the components of A in cartesian (expressed in x, y, and z), cylindrical (ex-
pressed using , and z) and spherical coordinates (expressed in r, , and ). In all cases a, b, and c are
constants.
(a). A = a x +b y +c z
A is constant because any point in the eld will give you the same unit vector at that point.
The Divergence of A is given by
A =
A
x
+
A
y
+
A
z
=
a
x
+
b
y
+
c
z
= 0
3
The curl of A is given by
A =
x y z
z
a b c
= 0
in cartesian coordinates the components of A are given by
A
x
= a A
y
= b A
z
= c
in cylindrical coordinates we must use a transformation matrix to go from one basis to another, i.e
cos sin 0
sin cos 0
0 0 1
x
y
z
where the 3x3 matrix is the transformation matrix that lets you represent cylindrical coordinates in
cartesian coordinates. Thus we nd
= cos x+sin y
= sin x+cos y
z = z
thus the components are given by
A
= A = acos+bsin
A
=

A =asin+bcos
A
z
= z A = c
In spherical coordinates we must also use a transformation matrix.
sincos sinsin cos

coscos cossin sin
sin cos 0
x
y
z
where the 3x3 matrix is the transformation matrix that lets you represent the spherical coordinates in
cartesian coordinates, this yields
r = sincos x+sinsin y+cos z
= coscos x+cossin ysin z
= sin x+cos y
thus we nd the components to be given by
A
r
= r A = asincos+bsinsin+ccos
A
=

A = acoscos+bcossincsin
A
=

A =asin+bcos
\
4
(b). A = a +b
+c z
A is not constant because as your position changes, so do and

. The component is the distance
from the z axis and the direction is always in the direction.
A =
1
(a) +
1
+
c
z
=
a

A =
b
z
in cartesian coordinates we must use
x
y
z
cos sin 0
sin cos 0
0 0 1
where the 3x3 matrix is the transformation matrix that lets you represent cylindrical coordinates in
cartesian coordinates. Thus we nd
x = cos sin
y = sin +cos
z = z
thus the components are given by
A
x
= x A = acosbsin
A
y
= y A = asin+bcos
A
z
= z A = c
in cylindrical coordinates the components of A are given by
A
= a A
= b A
z
= c
In spherical coordinates we must also use a transformation matrix.
sin 0 cos
cos 0 sin
0 1 0
cylindrical coordinates, this yields
r = sin +cos z
= cos sin z

A
r
= r A = asin+ccos
A
=

A =acoscsin
A
=

A = b
5
(c). A = a r +b
+c
A is not constant with the same argument as above except that r,

, and all change according to where
the point is.
A =
1
r
2
l
(r
2
a) +
1
r sin
(sinb) +
1
r sin
c
=
2a
r
+
c
r
cot
A =
1
r sin
(sinc)
b
r +
1
r
1
sin
a

r
(rc)
+
1
r

r
(rb)
a
=
c
r
cot r
c
r
+
b
r
in cartesian coordinates the components of A are given by
x
y
z
sincos coscos sin

sinsin cossin cos
cos sin 0
cartesian coordinates, this yields
x = sincos r +cos cos
+sin
y = sinsin r +cossin
cos
z = cos r +sin

A
x
= x A = asincos+coscos+csin
A
y
= y A = asinsin+bcossinccos
A
z
= z A = acosbsin
in cylindrical coordinates the components of A are given by
sin cos 0
0 0 1
cos sin 0
cylindrical coordinates, this yields
= sin r +cos
z = cos r sin
6
A
= A = asin+bcos
A
=

A = c
A
z
= z A = acosbsin
and in spherical coordinates they are given by
A
r
= a A
= b A
= c
Problem # 3
Show that
Z
V
f dV =
I
A
f dA
where A is the area of the closed surface bounding the volume V. Hint: multiply each side by a constant
vector and use calculus theorems.
If we multiply both sides by a constant vector eld A we get
A
Z
V
f dV = A
I
A
f da
if we work on the left-hand side, along with using equation 5 from the front of Grifths we nd
Z
V
(A f )dV =
Z
V
( f A)dV
Z
V
f ( A)dV
we know that the right hand term goes to zero because the gradient of a constant vector eld is 0.
Invoking the Divergence theorem on the left hand term we get
Z
V
( f A)dV =
I
A
f A da
we nally nd
A
Z
V
f dV =
I
A
f A da
A
Z
V
f dV =
I
A
A ( f da)
A
Z
V
f dV = A
I
A
f da
thus we have
Z
V
f dV =
I
A
f dA
Problem # 4
Show that the velocity (dr/dt = r) is expressed in cylindrical coordinates as +
+ z z and in
spherical coordinates as r r +r
+r sin
.
7
We know that the velocity can also be written as
r =
dx
dt
x +
dy
dt
y +
dz
dt
z
and we know that in cylindrical coordinates
x = cos
d
dt
(cos) = cos
sin
y = sin
d
dt
(sin) = sin+

cos
z = z
d
dt
(z) = z
and we also know that in cylindrical cooridinates
x = cos sin
y = sin +cos
z = z
thus we can write the velocity as
r = ( cos
sin)(cos sin
) +( sin+

cos)(sin +cos
) + z z
= cos
2
+

sin
2
+ sin
2
+

cos
2
+ z z
= +
+ z z
in Spherical coordinates we can apply the same technique.and we know that in spherical coordinates
x = r sincos
d
dt
(r sincos) = r sincos+

r coscos
r sinsin
y = r sinsin
d
dt
(r sinsin) = r sinsin+

r cossin+

r sincos
z = r cos
d
dt
(r cos) = r cos
sin
and we also know that in cylindrical cooridinates
x = cos sin
y = sin +cos
z = z
thus we can write the velocity as being (minus some algebra)
r =
dx
dt
x +
dy
dt
y +
dz
dt
z = r r +r
+r sin
Problem # 5
Here we will practice the use of eld differential operators and visualize some elds.
(a). Find the gradient of the scalar potential (x, y, z) = xy. Provide a clear sketch of the countour
lines of in the x y plane and a representation of its gradient eld. Such a eld is known as a radial
8
quadrupole eld, and is used in the focusing of charged particle beams and of dipolar (electric or magnetic)
particles.
The gradient of is given by
=

x
(xy) +

y
(xy) +

z
(xy)
= (y x +x

y)
The sketch is given as
(b). Provide a clear sketch (in the x y plane) of the vector eld expressed in cylindrical coordinates as
E = where is a constant. Calculate its divergence. If E is an electric eld, what charge distribution
generates it?
The gradient of E is given by
E =
(
2
)
= 2
The divergence of E is given by
E = 2
9
which is the direction of most change. The charge distribution that produces this charge can be found
by using Gausss law
E =
q
enc
0
= 2 q
enc
= 2
0
The sketch is given by
(c). Consider a large sheet placed on a turntable that is set rotating at angular velocity . What vector
eld v describes the local velocity on this sheet? Provide a clear sketch of this eld. What is the curl of
this vector eld (now expressed as a vector in a three-dimensional space, i.e including a coordinate for the
z axis)? If a magnetic vector potential A is dened with the functional form of v (just a different constant
out front), what magnetic eld B =A does it represent.
The vector describing this system is given by
v =r
where r is the distance from the origin and is the angle between any lines originating from the origin.
The curl is given by
v =
1
r
r
(r
2
) z = 2 z
The magnetic eld is given by
B =v = 2 z
The sketch is given by
10
(d). Now generate new vector eld by taking your vector eld v obtained in part (c) above, and adding
(1) a constant eld, and (2) a radial quadrupole eld from part (a). Provide clear sketches of these two
elds and compare with that of v. Show explicitly that the curl of these three elds is the same. Note that
these three elds all represent forms of a magnetic vector potential that describes the same magnetic eld.
We can dene a constant vector eld as
k = a x+b y+c z
which can also be represented as
k = (acos+bsin) +(asin+bcos)
+c z
so now we can add these two vectors
v+k = (acos+bsin) +(asin+bcos+r)
+c z
From part (a) we found
= (y x+x y)
= (ycos+xsin) +(xcosysin)
adding up these vectors yield

v+ = (ycos+xsin) +[r+(xcosysin)
To nd the curl we use

(v+k) =v+k
the right hand term goes to 0 and the left hand term is
v = 2 z
11
also
(v+) =v+
the right hand term is also 0, so we get
v =2 z
we have just showed explicitly that all the curls are the same,. For the rs sketch we can dene a
constant vector eld as
k = u y
and then add these two vector elds, as show in the diagram part (a), the second diagram (b) is the sum
of our original vector eld added to the quadrupole eld given by (a).
Problem # 6
(a) Show that
Z
s
f (A) da =
Z
s
[A(f )] da+
I
p
f A dl
Using equation 7 from the front cover of Grifthss we nd
Z
s
f (A) da =
Z
s
[( f A)] da+
Z
s
[A(f )] da
invoking Stokes theorem on the left hand term yield
Z
s
[( f A)] da =
I
p
f A dl
thus we nd
Z
s
f (A) da =
Z
s
[A(f )] da+
I
p
f A dl
12
(b) Show that
Z
V
B (A)d =
Z
V
A(B)d +
I
s
(AB) da
Using equation 6 from thr front cover of Grifths we nd
Z
V
B (A)d =
Z
V
(AB)d +
Z
V
A (B)d
Using Gausss theorem on the left hand term we nd
Z
V
(AB)d =
I
s
(AB) da
Thus we have
Z
V
B (A)d =
Z
V
A(B)d +
I
s
(AB) da
Problem # 7
(a) Let F
1
=x
2
z and F
2
= x x+y y+z z. Calculate the divergence and curl of F
1
and F
2
. Which one can
be written as a gradient of a scalar? Find a scalr potential that does the job. Which one can be written as a
curl of a vector? Find a suitable vector potential.
The divergence of F
1
is given by
F
1
=

z
x
2
= 0
and the curl is given by
F
1
=
x y z
z
0 0 x
2
=

x
x
2
y =2x y
The divergence of F
2
is given by
F
2
=

x
x +

zy
y +

z
z = 3
and the curl is given by
F
2
=
x y z
z
x y z
= 0
We knowthat F
1
can be written as the curl of a vector potential because it is divergen-less or solenoidal,
to nd the vector potential we do
F
1
= V
We need to nd the vector potential V, so we can nd it using
F
1
=V =
x y z
z
A
x
A
y
A
z
=
A
y
x
z = x
2
z
13
thus A
y
is given by
A
y
=
x
3
3
and the vector potential is given as
V = a x+
x
3
3
y+b z
where a and b are constants. We know that F
2
can be written as the gradient of a scalar potential
because it is curl-less. we nd the scalar potential by using
F
2
= V
=
x
2
2
+
y
2
2
+
z
2
2
thus we know that the scalar potential is

V =
x
2
2
+
y
2
2
+
z
2
2
(b) Show that F

3
= yz x+zx y +xy z can be written both as the gradient of a scalar and as the curl of a
vector. Find scalar and vector potentials for this function.
First we must show that F
3
is is both divergen-less and curl-less, to show divergen-less and nd the
vector potential we use
F
3
=

x
(yz) +

y
(xz) +

z
(xy) = 0
thus we must use
F
3
= V =
x y z
z
A
x
A
y
A
z
A
z
y

A
y
z
A
z
x

A
x
z
y+
A
y
x

A
x
y
z
Solving for A
x,
, A
y
, and A
z
we nd
yz x =
A
z
y

A
y
z
x A
z
=
zy
2
2
zx y =
A
x
x

A
z
z
y A
x
=
yz
2
2
xy z =
A
y
x

A
x
y
z A
y
=
xy
2
2
Thus the vector potential is
V =
zy
2
2
x+
xy
2
2
y+
zy
2
2
z
14
For the scalar potential we do
F
3
=
x y z
z
yz zx xy
= 0
by inspection. The scalar potential is given by
F
3
= yz x+zx y+xy z =V
and we nd
V =xyz
thus the vector can be written as
F
3
= (xyz)
Problem # 8
Check the divergence theorem for the function
v = r
2
sin r +4r
2
cos
+r
2
tan
using the volume of the ice-cream cone shown on page 56 Fig. 1.52 (the top surface is spherical,
with radius R and centered at the origin).
The divergence theorem states
Z
V
vdV =
I
A
v da
the left hand side can be solved by taking a volume integral of v.
v =
1
r
2
r
(r
4
sin) +
1
r sin
(4r
2
sincos) +
1
r sin
(r
2
tan)
= 4r sin+
4r
sin
(cos)
2
(sin)
2
= 4r sin+4r
1
sin
2(sin)
= 4r
(cos)
2
sin
and the integral is given by
Z
V
vdV =
Z
2
0
d
Z
R
0
4r
3
dr
Z
/6
0
(cos)
2
d
which yields
Z
V
vdV = 2R
4

12
+
3
8
Now for the area integral must look at the the cone and the surface separately
I
A
v da =
Z
2
0
R
4
d
Z
/6
0
(sin)
2
d+
1
2
Z
2
0
d
Z
R
0
sin(/6)cos(/6)4r
3
dr
= 2R
4

12
3
8
+R
4
3
4
= 2R
4

12
+
3
8
15
Problem # 9
A little more practice on vector analysis: consider the vector function F(r) =

where

is dened as
conventional for spherical coordinates.
(a). Calculate the line integral
I
c
F dl
where C is a circle of radius in the x y plane, centered at the origin, and where the integral is
evaluated with dl oriented counter clockwise.
from the diagram we can see that
dl =d
given that F(r) =

we nd
I
c
F dl =
Z
2
0
(d
) =
Z
2
0
d = 2
(b). Calculate the surface integral
Z
H
(F) da
where the surface H is a hemisphere that is above and bounded by the curve C used for part (a). The
surface integral is calculated with da oriented outward.
da =
2
sindd r
and to nd what the curl of F we do
F =
1
r sin
(sin)
1
r
r
(r)
=
cot
r
r
1
r
in our case r = thus we have

F =
1
[cot r
]
thus the integral is given by
Z
H
(F) da =
Z
H
1
[cot r
]
2
sindd r =
Z
/2
0
cosd
Z
2
0
d = 2
(c). Calculate the surface integral
Z
D
(F) da
where the surface D is now the disk in the x y plane that is bounded by the curve C used for part (a).
The surface integral is calculated with da oriented upward.
da =drd z = drd(cos r sin
) =

2
and to nd what the curl is we do
F =
1
r sin
(sin)
1
r
r
(r)
=
cot
r
r
1
r
16
in our case r = thus we have
F =
1
[cot r
]
thus the integral is given by
Z
D
(F) da =
Z
D
1
[cot r
] drd(cos r sin
)
=
Z
2
0
Z

0
cos
2
+sin
2
sin
drd
=
Z
2
0
d
Z

0
dr = 2
(d). Verify that Stokes theorem holds for both surfaces H and D.
Stokes theorem states
I
F dl =
Z
S
(F) da = 2
thus Stokes theorem holds.
Problem # 10
Let H(r) = x
2
y x+y
2
z y+z
2
x z. Find an irrotational function F(r) and a sonisoidal function G(r) such
that H = F+G.
We know that an irrotational function can be described as
F = 0
and a sonisoidal function is described as
G = 0
thus we know that
H =F+G = G
and we know that
H =
x y z
z
x
2
y y
2
z z
2
x
=y
2
xz
2
yx
2
z
thus we can nd G by
G =
x y z
z
G
x
G
y
G
z
G
z
y

G
y
z
G
z
x

G
x
z
y+
G
y
x

G
x
y
z
looking at the following terms we nd
G
z
y
=y
2
G
x
z
=z
2
G
y
x
=x
2
17
thus
G
z
=
y
3
3
G
x
=
z
3
3
G
y
=
x
3
3
thus we nd G to be
G =
z
3
3
x
x
3
3
y
y
3
3
z
since we know what H and G are we can easily nd F
F = HG =
x
2
y +
z
3
3
x+
y
2
z +
x
3
3
y+
z
2
x +
y
3
3
z
and it is seen that
F = 0
G = 0
and by construction we know that
H = F+G
18
Chapter 2
Electrostatics
2.1 The Electric Field
2.1.1 Coulombs Law
The fundamental problem electromagnetic theory hopes to solve is this: We have some electric charges
q
1
, q
2
, q
3
,... (call them source charges); what force do they exert on another charge, Q (call it the test
charge). The force on the test charge Q due to a single point charge q which is at rest a distance r
away
is given by Coulombs Law
F =
1
4
0
qQ
r
2

r
(2.1)
The cosntant
0
is called the permitivity of free space. In SI units, where force is in Newtons (N), distance
in meters (m), and charge in Coulombs (C)
0
= 8.8510
12
C
2
N m
2
In words, the force is proportional to the product of the charges and inversely proportional to the square
of the seperation distance. As always
r
= r r
where r
is the seperation vector from the location of q to Q, r is the magnitude and

r
is the direction.
The force points along the line from q to Q; it is repulsive if q and Q have the same sign, and attractive if
their signs are different. Coulombs law and the principle of superposition constitute the physical input for
electrostatics- the rest, except for some special properties of matter, is mathematically elaborationof these
fundemental rules.
2.1.2 The Electric Field
If we have several point charges q
1
, q
2
, ..., q
n
at distances r
1
, r
2
, ..., r
n
from Q, the total force on Q is
evidently
F =F
1
+F
2
+... =
1
4
0
q
1
Q
r
2
1
r
1
+
q
2
Q
r
2
2
r
2
+..
or
F = QE
19
where
E(r)
1
4
0
n
i=1
q
i
r
2
i
r
i
(2.2)
E is called the electric eld of the source charges.
2.1.3 Continuous Charge Distributions
Our denition of the electric eld Equation 2.2 assumes that the source of the eld is a collection of
discrete point charges q
i
. If, instead, the charge is distributed continuously over some region, the sum
becomes an integral
E(r) =
1
4
0
Z
1
r
2
rdq
If the charge is spread out along a line, with charge-per-unit length , then dq = dl
(where dl
is an
element of length along the line); if the charge is smeared out over a surface, with charge-per-unit are
, the dq = da
, (where da
is an element of area on the surface) and if the charge lls a volume, with
charge-per-unit volume then dq = d
(where d
is an element of volume)
dq dl
da
thus the electric eld of a line charge is

E(r) =
1
4
0
Z
(r
)
r
2
rdl
for a surface charge

E(r) =
1
4
0
Z
(r
)
r
2
rda
and for a volume charge

E(r) =
1
4
0
Z
(r
)
r
2
rd
(2.3)
Equation 2.3 itself is often referred to as Coulombs law, because it is such a short step from the original,
and because a volume charge is in a sense the most general and realistic case.
2.2 Divergence and Curl of Electrostatic Fields
2.2.1 Field lines, Flux, and Gausss Law
The ux of E through a surface S is given as
E

Z
S
E da
it is a measure of the number of eld lines passing through S. This suggests that the ux through any
closed surface is a measure of the total charge inside. This is in essence Gausss Law. In the case of a
point charge q at the origin, the ux E through a sphere of radius r is
I
E da =
Z
1
4
0
q
r
2
r
(r
2
sindd r =
q
0
20
For any closed surface, then
I
E da =
1
0
Q
enc
where Q
enc
is the total charge enclosed within the surface. As it stands, Gausss law is an integral
equation, but we can readily turn it into a differential one, by applying the divergence theorem
I
S
E da =
Z
V
( E)d
rewriting Q
enc
in terms of the charge density , we have
Q
enc
=
Z
V
d
so Gausss law becomes
Z
V
( E)d =
Z
V
d
and since this holds for any volume the integrand must be equal
E =

0
(2.4)
Equation 2.4 is Gausss law in differential form.
2.2.2 The Curl of E
Ill calculate the curl of E, as I did the divergence by studying the simplest possible conguration, a point
charge at the origin. In this case
E =
1
4
0
q
r
2
r
what if we calculate the line integral of this eld from some point a to some other point b
Z
b
a
E dl
in spherical coordinates, dl = dr r +rd
+r sind
, so
E dl =
1
4
0
1
r
2
dr
therefore
Z
b
a
E dl =
1
4
0
q
r
a
q
r
b
where r
a
is the distance from the origin to point a and r
b
is the distance to point b. The integral around a
closed path is evidently zero (for then r
a
= r
b
)
I
E dl = 0
and hence applying Stokes theorem
E = 0 (2.5)
Equation 2.5 holds for any static charge distribution.
21
2.3 Electric Potential
The electric eld E is not just any old vector; it is a special kind of vector function, one whose curl is
always zero. There is a theorem that asserts that any vector whose curl is zero is equal to the gradient of
some scalar. Because the line integral is indepenent of path, we can dene a function
V(r)
Z
r
O
E dl (2.6)
this is called the electric potential. This allows us to write the electric eld as
E =V (2.7)
Equation 2.7 is the differential version of Equation 2.6, it says that the electric eld is the gradient of a
scalar potential. If you knowV, you can easily get E by just taking the gradient of V.
2.3.1 Poissons Equation and Laplaces Equation
We found that the electric eld can be written as the gradient of a scalar potential. The question arises:
What do the fundamental equations for E,
E =

0
E = 0
look like, in terms of V? Well. E = (V) =
2
V, so apart from the persisting minus sign, the
divergence of E is the Laplacian of V. Gausss law then says
2
V =
0
(2.8)
this is known as Poissons equation. In regions where there is no charge, so that =0, Poissons equation
reduces to Laplaces equation
2
V = 0
2.3.2 The Potential of a Localized Charge Distribution
I dened V in terms of E. Ordinarily, though it is E that were looking for. The idea is that it might be
easier to get V rst, and the calculate E by taking the gradient. In general, the potential of a point charge
q is
V(r) =
1
4
0
q
r
or for continuous distribution
V(r) =
1
4
0
Z
1
r
dq
in particular, for a volume charge
V(r) =
1
4
0
Z
(r
)
r
d
this is the equation we are looking for, telling us how to compute V when we know ; it is, if you like, the
solution to Poissons equation, for a localized charge distribution.
22
2.3.3 Summary; Electrostatic Boundary Conditions
The electric eld always undergoes a discontinuity when you cross a surface charge . In fact, it is a
simple matter to nd the amount by which E changes at such a boundary. Gausss law states that
I
S
E da =
1
0
Q
enc
=
1
0
A
where A is the area of a pillbox used on the surface. Now the sides of the pillbox contribute nothing to the
ux, in the limit as the thickness goes to zero, so we are left with
E
above
E
below
=

0
Conclusion: The normal component of E is discontinuous by an amount /
0
at any boundary. In par-
ticular, where there is no surface charge, E
is contnuous, as for instance at the surface of a uniformly

charged solid sphere. The tangential component of E, by contrast, is always contnuous.
E
above
= E
below
the boundary conditions on E can be combined into a single formula
E
above
E
below
=

0
n
where n is a unit vector perpendicular to the surface, pointing from below to above. The potential,
meanwhile, is continuous across any boundary, since
V
above
V
below
=
Z
b
a
E dl
as the path length shrinks to zero, so too does the integral
V
above
=V
below
However, the gradient of V inherits the discontinuity in E; since E =V, implies
V
above
V
below
=
0
n
or more conveniently
V
above
n

V
below
n
=
0
where
V
n
= V n
denotes the normal derivative of V (that is, the rate of change in the direction perpendicular to the
surface).
23
2.4 Work and Energy in Electrostatics
2.4.1 The Work Done to Move a Charge
Suppose you have a stationary conguration of source charges, and you want to move a test charge Q from
point a to point b. How much work will you have to do? The work is dened as
W =
Z
b
a
F dl =Q
Z
b
a
E dl = Q(V(b) V(a)]
notice that the answer is independent of the path you take from a to b; in mechanics, then, we would call
the electric force conservative. Dividing through by Q, we have
W
Q
=V(b) V(a)
in words, the potential difference between two points is equal to the work-per-unit charge required to carry
the particle between the points. In particular, if you want to bring the charge Q in from far away and stick
it at a point r, the work you must do is
W = Q[V(r) V()]
so if you have set the reference point out at innity
W = QV(r)
2.4.2 The Energy of a Point Charge Distribution
How much work would it take to assemble an entire collection of point charges? Imagine bringing in the
charges, one by one, from far away. The work needed to assemble a conguration of point charges is given
by
W =
1
2
n
i=0
q
i
V(r
i
) (2.9)
this is also the amount of work youd get back out if you dismantled the system. In the mean time it
represents the energy stored in the conguration.
2.4.3 The Energy of a Continuous Charge Distribution
For a volume charge density , Equation 2.9 becomes
W =
1
2
Z
Vd
(The corresponding integrals for line and surface charges would be
R
Vdl and
R
Vda, respectively. This
can also be written as
W =

0
2
Z
space
E
2
d
The energy can be thought of as being stored in the eld or stored in the charge. We can also say
0
2
E
2
= energy per unit volume
24
Because electrostatic energy is quadratic in the eld, it does not obey a superposition principle. The energy
of a compound system is not the sum of the energies of its parts considered seperately, there are also cross
terms
W
tot
=W
1
+W
2
+
0
Z
E
1
E
2
d
for example, if you double the charge everywhere, you quadruple the total energy.
2.5 Conductors
2.5.1 Basic Properties
In an insulator, such as glass or rubber, each electron is attached to a particular atom. In a metallic
conductor, by contrats, one or more electrons per atom are free to roam about at will through the material.
A perfect conductor would be a material containing an unlimited supply of completely free charges. There
are some general rules when dealing with conductors
E=0 inside a conductor
= 0 inside a conductor
any net charge resides on the surface
A conductor is an equipotential
E is perpendicular to the surface, just outside a conductor
2.5.2 Surface Charge and the Force on a Conductor
Because the eld inside a conductor is zero, boundary conditions requires thatb the eld immediately
outside is
E =

0
n
consistent with our earlier conclusion that the eld is normal to the surface. In terms of potential
=
0
V
n
these equations enable you to calculate the surface charge on a conductor, if you can determine E or V. In
the presence of an electric eld, a surface charge will, naturally experience a force; the force per unit area,
f, is E. But theres a problem here, for the electric eld is discontinuous at a surface charge, so which
value are we supposed to use? The answer is that we should use the average
f = E
average
=
1
2
(E
above
+E
below
)
the force per unit area is also given as
f =
1
2
0
2
n
this amounts to an outward electrostatic pressure on the surface, tending to draw the conductor into the
eld, regardless of the sign of . Expressing the pressure in terms of the eld just outside the surface
P =

0
2
E
2
whic is identical to the energy-per-unit volume stored in the elds.
25
2.5.3 Capacitors
Suppose we have two conductors, and we put charge +Q on one and Q the other. Since V is constant
over a conductor, we can speak unambiguously of the potential difference between them
V =V
+
V
=
Z
+
E dl
We dont know how the charge distributes itself over the two conductors, and calculating the eld would
be a mess, if their shapes are complicated, but this much we know: E is proprtional to Q. Since E is pro-
portional to Q, so also is V. The constant of proportionality is called the capacitance of the arrangement:
C =
Q
V
Capacitance is a purely geometrical quantity, determined by the sizes, shapes, and seperation of the two
conductors.
Problem # 1
Electrostatics with cylindrical symmetry
(a). Find the electric potential V(z) a distance z above the center of a circular loop of radius R which
carries linear charge density . Obtain an expression correct for all values of z (not just z > 0).
We know that
V =
1
4
0
Z
dq
r
since we know that dq is given by
dq = ds = Rd r =
R
2
+z
2
thus the above expression for the potential is
V =
1
4
0
Z
2
0
Rd
R
2
+z
2
=

2
0
R
R
2
+z
2
(b). Find the electric potential a distance z above the center of a circular disk of radius R, which carries
a uniform surface charge .
26
We know that
dq = da = 2rdr
putting this into the expression for the potential we nd
V =
1
4
0
Z
dq
r
=V =
1
4
0
Z
R
0
2r
z
2
+r
2
dr
using a u substitution we nd the solution to the integral to which is given by
u = z
2
+r
2
du = 2rdr (r = 0 : u z
2
, r = R : u z
2
+R
2
)
which gives us
V =

4
0
Z
z
2
+R
2
z
2
u
1/2
du =

2
0
z
2
+R
2
z
(b). Using the potential and the symmetry of the problem, determine the electric eld a distance z
above the center of the circular disk for part b. Check that your answer makes sense: What do you get
for |z| R and for z R: Show also that he magnitude of the eld at z is proprtional to the solid angle
subtended by the disk as seen from the point z.
We know that the electric eld can be dened as
E =V =

2
0
z
2
+R
2
z
2
z =

2
0
z
2
+R
2
1
z
For R z we can use a Taylor expansion
(1+x)
n
1+nx
so we nd
E =

2
0
z
2
+R
2
1
z =

2
1+
R
2
z
2
1/2
1
z

4
0
R
2
z
2
z
and we know that
q =R
2
z = R
thus at R z we should see a point charge i.e
E =
q
4
0
R
2
z
27
For R z we can use a Taylor expansion
(1+x)
n
1+nx
so we nd
E =

2
0
z
2
+R
2
1
z =
z
R
1+
z
2
R
2
1/2
1

2
z

2
0
1
2
z
3
R
3
z
R
+1
z
this is only for R z.
To show also that the magnitude of the eld at z is proprtional to the solid angle subtended by the disk
as seen from the point z. We will say
=
Z
2
0
d
Z

0
sind =2cos
but the limits of are given by
0 and = cos
1
z
2
+R
2
which gives us
= 2z
z
2
+R
2
1
and
E
=

4
0
= constant
thus
E = k
Problem # 2
Two spheres, each of radius R carrying uniform charge densitys + and , respectively, are placed
so that they partialy overlap. Call the vector from the positive center to the negative center d. Show that
the eld in the region of overlap is constant, and nd its value.
We can solve this problem by nding the electric elds for 1 sphere and adding the spheres using
superposition. It is well known that the electric eld of a uniform sphere is given by
I
E da =
q
tot
0
q
tot
=(r)V =
4
3
(r)r
3
since the electric eld is constant we nd
EA =
4
3
(r)r
3
1
0
E =
(r)r
3
0
r
so we nd that the two spheres are given by
E
+
=
(r)r
+
3
0
r E
=
(r)r
3
0
r
28
and so the total electric eld is just the sum of these two (superposition), we also know that the vectors
are pointing towards each other and that the vertical components of the vectors cancel by symmetry. We
can nd the total electric eld as
E
t
= E
+
+E
(r)r
+
3
0
(r)r
3
0
r =
(r)
3
0
[r
+
r
] r
but since we know that the total distance between this two vectors will always be the same we can
write the r
vector as r
which would give you just give you d (which is constant) and thus your total
electric eld would be constant. i.e
d =r
+
+r
= r
+
r
Problem # 3
Consider the so-called screened Coulomb potential of a point charge q that arises, for example, in
plasma physics
V(r) =
q
4
0
e
r/
r
where is a constant (called the screening length)
(a). Determine the charge distribution (r) that produces this potential. Sketch this function in a
manner that clearly describes all of its characteristics (i.e whats the best way of representing this three-
dimensional charge distribution? Use it and explain what your potting.
We can solve this problemtwo ways, you can take the Laplacian of a Laplacian or just use the following
identities
2
V = (( f g)) = ( f g+gf )
and we also know the following relationships
E =V E = (V) =
(r)
0
thus we know that
(r) =
0
( E) =
0
( (V))
so we need to nd what the divergence of the electric eld is, by nding the gradient of the potential
rst and taking the divergence of this
V =

r
q
4
0
e
r/
r
=
q
4
0
e
r/
r
=
e
r/
r
2
1+
r
r
thus we nd
E =
e
r/
r
2
1+
r
r
letting
f = e
r/
1+
r
and g =
r
r
2
we nd
E = e
r/
1+
r
r
r
2
+
r
r
2
e
r/
1+
r
29
but we know from Grifth page 50 that
r
r
2
= 4
3
(r)
so we nd
E = e
r/
1+
r
4
3
(r) +
r
r
2
e
r/
1+
r
= e
r/
1+
r
4
3
(r) +
1
r
2
e
r/
1+
r
=
q
4
0
e
r/
1+
r
4
3
(r)
e
r/
r
2
thus the charge density is given by

(r) =
q
4
e
r/
1+
r
4
3
(r)
e
r/
r
2
The plot is given by

where the plot is a countour map that describes the distribution with respect to the radius.
(b). Show by explicit calculation over (r) that the net charge represented by this distribution is zero.
Explain how this observation is easily derived using the integral form of Gauss law.
the total charge Q can be found by
Q
net
=
Z
V
(r)d =
q
4
Z
space
e
r/
1+
r
4
3
(r)d
Z
space
e
r/
r
2
d
which gives us
Z
V
(r)d =
q
4
44
Z

0
re
r/
2
dr
where the term on the right hand side is just the Gamma function and the delta function integral is just
4 by denition. So we nd that the total charge is zero
Z

0
re
r/
2
dr = 1
30
Q(tot) =
Z
space
(r)d =
q
4
[44] = 0
this tedious calculation could have been solved by using the divergence theorem and Gausss law. i.e
Z
V
( E)dV =
I
s
E da =
q
enc
0
but we know t
Z
V
( E)dV =
Z
(r)
dV = 0 =
I
E da =
q
enc
0
Since we now know that we have a +q charge due to the delta function and a -q charge from the integral
of the potential contained within a surface S then the surface has the same number of eld lines coming
in that are going out and thus the surface integral goes to zero because the divergence goes to zero thus
implying that q
enc
is zero.
Problem # 4
(a). Show that the electrostatic energy of a uniformly charged solid sphere, with total charge Q and
radius R is 3Q
2
/(20
0
R).
It is assumed that we know what the electric eld is in the inside of the sphere and on the outside of
the sphere is. i.e
E
in
=
q
4
0
R
3
r r E
out
=
q
4
0
r
2
r
and using the following expression for the electrostatic energy
W =

0
2
Z
space
E
2
d
thus we must consider two integrals, the contribution from the inside and the contribution from the
outside. Knowing the limits of integration we nd
W =
q
2
8
0
Z
R
0
r
4
R
6
dr +
Z

R
1
r
2
dr
=
q
2
8
0
1
5R
+
1
R
W =
3q
2
20
0
R
(b). Use the result above to compute the electrostatic energy of an atomic nucleus with Z protons and a
total of A nucleons, using an approximation for the nuclear radius as R = (1.210
15
m)A
1/3
. Give your
results in units of MeV times Z
2
/A
1/3
this is just a plug and chug problem, gure out your constants and plug away, i.e
W =
3(Zq)
2
20
0
(1.210
15
m)A
1/3
0.720MeV
Z
2
A
1/3
(c). Calculate the change of electrostatic energy when a uranium nucleus undergoes ssion.
31
We know that U
235
decays as
U
235
92
36
Kr +
141
56
Ba+3n
thus we know that
W =W
reac
W
u
=|(W
kr
+W
B
) W
u
thus
W = 0.720MeV
36
2
92
1/3
+
56
2
141
1/3
92
2
235
1/3
=347.02MeV
which is a large amount of liberated energy.
Problem # 5
Evaluate the following integrals
(a).
Z
space
(r
2
+r a+a
2
)
3
(r a)d
where a is a xed vector and a is its magnitude.
We know that the delta function is zero everywhere except at r = a thus we must evaluate this integral
at a i.e
Z
f (r)
3
(r a)d = f (a)
thus this yields
Z
space
(r
2
+r a+a
2
)
3
(r a)d = 3a
2
(b).
Z
V
|r b|
2
3
(5r)d
where V is a cube of side 2, centered on the origin, and b = 4 y+3 z.
We know that we must evaluate this function at r = 0, which yields
Z
V
|r b|
2
3
(5r)d =| b|
2
= 25
(c).
Z
V
(r
4
+r
2
(r c) +c
4
)
3
(r c)d
where V is a sphere of radius 6 about the origin, and c= 5 x+3 y+2 z.
We know that the delta function is zero everywhere excpet for where it is dened, so this means if the
limits of the integral are not within the specied value at which the delta function is equal to 1 then the
integral is zero.
c =
5
2
+3
2
+2
2
=
38 = 6.16 6.16 > 6.0

thus this implies that the magnitude of c is outside of the range and so
Z
V
(r
4
+r
2
(r c) +c
4
)
3
(r c)d = 0
32
(d).
Z
V
r(dr)
3
(e r)d
where d = (1, 2, 3),and e = (3, 2, 1), and V is a sphere of radius 1.5 centered at (2,2,2).
From our experience with delta function we now know that this will be given by
Z
V
r(dr)
3
(e r)d = e (de) = 4
Problem # 6
The farad is actually an enormous unit of capacitance. To illustrate this, treat the Earth as a conducting
sphere and nd its capacitance.
+
+Q
R2
R1
E
E
E
Q
+
The above gure describes the earth capacitance problem, Where R
1
is the radius of the earth and also
R
2
R
1
and also a positive charge Q on the outside surface and a negative charge Q for the earth.
We know that the capacitance is dened as
C =
Q
V
where Q is the charge and V is the potential. The potential is given as
V =
Z
R
2
R
1
E dr =
1
4
0
Z
R
2
R
1
Q
r
2
(cos)dr =
1
4
0
1
R
2
1
R
1
because the electric eld points in the r direction we must consider the cos term. Thus we nd
V =
Q
4
0
1
R
2
1
R
1
and so the capacitance is given by

C =
R
1
R
2
4
0
R
2
R
1
R
2
R
1
R
1
= 6, 378 km
33
so we nd
C = R
1
4
0
= 700 F
Problem # 7
Consider a system of n conductors that are prepared with total charges of Q
i
, with i{1, 2, 3, . . .
n} labeling the conductors. The surface charge densities on each of the conductors are described by the
functions
i
, for which, clearly
Q
i
=
Z
S
i
i
da
i
with the integral being over the surface S
i
of conductor i.
(a). Find an expression for the potential on conductor i. This you may obtain as usual by summing
over the contributions to the potential from all the surface charges on all the conductors (including the
i th conductor).
Since we know that there i
th
and j
th
conductors, to nd the potential of the i
th
conducter from all the
j
th
conductors, we must sum over all the j
th
conducters contribution to the potential, i.e
V
i
=
1
4
0
Z
S
j
Q
j
r
i j
=

j=1
Z
S
j
j
da
j
r
i j
where r
i j
is the distance from the i
th
conducter to the j
th
conductor.
(b). Let us dene a coefcient p
i j
as
p
i j
=
1
4
0
S
j
Z
S
j
f
i
da
j
r
i j
Here, r
i j
denotes the distance from one point on conductor i to the point on conductor j at which the
integrand is evaluated (as we integrate over the surface of j ). In this integral, we make us of a relation
i
=
j
f
j
=
Q
j
S
j
f
j
in which S
j
is now the area of the j
th
conductor. The dimensionless function f
j
describes how the actual
charge density differs from the average. With this denition, show that the following set of equations hold:
V
1
= p
11
Q
1
+ p
12
Q
2
+... + p
1n
Q
n
V
2
= p
21
Q
1
+ p
22
Q
2
+... +P
2n
Q
n
.
.
V
n
= p
n1
Q
1
+ p
n2
Q
2
+... +P
nnQ
n
The coefcients p
i j
are known as coefcients of potential. They express how the potential on conductor
i will vary (linearly) upon changing the charge Q
j
on conductor j . We note here that the coefcients no
longer contain any reference to the specic potentials or charges placed on the conductors - they simply
reect geometric properties of the conductors and their placements.
(c). By considering the quantity p
i j
Q
i
Q
j
, show that p
i j
= p
ji
. This nice relation expresses the fol-
lowing fact: if a charge Q on conductor i brings that conductor to a potential V , then the same charge Q
placed on conductor i would bring conductor j to the same potential V.
34
using the fact that these matrix elements are symmetric, nn, and diagonalizable, we can use a linear
algebra by invoking the theory of matrix multiplication, i.e
n
i=1
(AB)
ii
=
n
i=1
n
j=1
A
i j
B
ji
=
n
j=1
n
i=1
B
ji
A
i j
=
n
j=1
(BA)
j j
this can be shown by
p
i j
Q
i
Q
j
=
Q
i
Q
j
4
0
S
j
Z
S
j
f
j
da
j
r
i j
=
Q
j
Q
i
4
0
S
i
Z
S
i
f
i
da
i
r
ji
= p
ji
Q
j
Q
i
this comes from the symmetry of the problem, i.e whatever conductor i sees will be exactly the same
if conductor i now became conductor j.
(d). Consider a situation with only two conductors that are used as a capacitor. What is the capacitance
for this system expressed in terms of the coefcients of potential p
11
, p
12
and p
22
?
if we know that
V
1
= p
11
Q
1
+ p
12
Q
2
V
2
= p
21
Q
1
+ p
22
Q
2
we also know that
p
12
= p
21
thus we can nd the potential by using
V =V
2
V
1
= (p
12
p
11
)Q
1
+(p
22
p
12
)Q
2
and we we assume that Q
1
=Q
2
we nd
C =
Q
1
V
=
1
p
11
+ p
12
+ p
22
Problem # 8
Suppose that a parallel plate ca-pacitor has rectangular plates but the plates are not exactly parallel.
The separation at one edge is d a and d +a at the other edge, where a d. Show that the capacitance is
given approximately by
C

0
A
d
1+
a
2
3d
2
where A is the area of the plate. A gure is given by

35
Thus we know from the diagram
A = 4lw
and knowing from Gausss law
I
E da =
dA
2
0
which give us
EdA =
dA
2
0
E =

2
0
( z)
to nd the potential we must use
V =
Z
E dl
and knowing that
dl = zds
where ds is in the direction of the eld. Thus we nd the potential to be given by
V =
Z

2
0
( z) ( zds) =

2
0
S
which S is the the distance between the plane that is a function of x which is given by the equation of
the line in the form
S = d +
a
w
x
thus we now know that the capacitance is given by and q
enc
= dA
C =
Z
w
w
2
0
dA
d +
a
w
x
=
Z
w
w
2
0
l
Z
w
w
dx
d +
a
w
using a u substitution, in which we will just qoute the solution. We know how to do u substitutions at
this level
C = 2
0
lw
a
ln
1+
a
d
ln
1
a
d
which if we do a bit of algebra with a knowledge of how the natural log functions work along with the
Taylor expansion for the natural log function which is
ln(x +1) x
x
2
2
+
x
3
3
...
we nd
C = 2
0
lw
a
a
2
a
2
2d
2
+
a
3
3d
3
a
d
a
2
2d
2
a
3
3d
3
the even terms cancel and thus to third order we nd, with a bit of algebra
C = 2lw
0
1
a
2a
d
+
a
3
3d
3
= 4lw
0
1
d
1+
a
2
3d
2
=
A
0
d
1+
a
2
3d
2
Problem # 9
(a) Two grounded conducting planes meet at an angle of = 60
0
at the origin. A point charge q is
a distance r
0
from the origin along their angular bisector. What image charges are needed to satisfy the
boundary conditions? (Hint: their arrangement is going to be highly symmetric)
36
We can see from the diagram will require 6 total charges and 5 image charges. This is necessary to
meet the boundary condition implied by the use of images. We can see that there is strong dependence in
the angle between the plates. You must balance the charges. where the total charge is zero.
What is the r depedence of the potential along the angular bisector for r r
0
?
We can nd the r dependence is given by
V =
q
4
0
1
r
1
1
r
2
1
r
3
+
1
r
4
+
1
r
5
1
r
6
if we pick an arbitrary point (0, 0, z) we can nd a relationship for the distance between these charges
and that point. The diagram is given by
knowing that the distance between each pair of charges is the same because it is a equilatiral triangle
and by symmetry we nd
r
4
= r
5
and also r
2
= r
3
and also from the diagram we can see
r
1
= r R
0
r
6
= r +R
0
Thus we nd that the potential is given by
V =
q
4
0
1
r R
0
2
r
2
+
2
r
4
1
r +R
0
37
combining the rst term and the last term gives us
V =
q
4
0
2R
0
r
2
+R
2
0
2
r
2
+
2
r
4
From the diagram we can also infer that

r
2
2
=
3
4
R
2
0
+
r
R
0
2
2
and by symmetry we know that
r
2
4
=
3
4
R
2
0
+
r +
R
0
2
2
thus we nd that the r dependence on the potential is given by
V =
q
4
0
2R
0
r
2
+R
2
0
3
4
R
2
0
+
r
R
0
2
2
+
2
3
4
R
2
0
+
r +
R
0
2
(b) Repeating the above problem but for different angles between the grounded plates. Show that if
= 2/n, where n is any even integer, then n 1 image charges are required to ground both sides of the
wedge. Show that if, however, n is an odd integer (e.g. n = 3) then an image solution cannot be obtained.
We can see that there is a dependence on n by
360
90
= 4 (41) = 3
360
180
= 2 (21) = 1
360
60
= 6 (61) = 5
from these three examples we can see that
2
= n total number of charges

and
n1 image charges
we can also see that must be even to allow an even number of charges that to be used in order to
satisfy the boundary conditions.
Problem # 10
A conductor is formed with a grounded conducting plane spanning the x y axis (z = 0). Upon it is
attached a hemispherical conductor of radius R
0
that extends into the half-plane z > 0 and whose center is
at the origin. A charge q is placed at the point (x, y, z) = (0, 0, z) with z > R
0
.
38
(a). What is the electrostatic potential V(r, ) with r and being conventionally dened spherical
coordinates (by symmetry, the potential must be independent of the azimuthal angle )? Hint: there will
be three image charges. Answer:
1
z
2
R
2
0
z
z
2
0
+R
2
0
A diagram is given by
Thus we know that
V(r, ) =
q
4
0
1
r
1
1
r
2
+
1
r
3
1
r
4
Using geometry and symmetry, along with the the example in Grifthss we nd that the relationships
for the all the radius are given as
r
1
2
= r
2
sin
2
+(r cosz)
2
r
2
2
= r
2
sin
2
+
r cos
R
2
0
z
2
and by symmetry we nd that the other two radius are given by
r
3
2
= r
2
sin
2
+
r cos+
R
2
0
z
2
r
4
2
= r
2
sin
2
+(r cos+z)
2
so we know that the potential is given by after cleaning it up a bit
V(r, ) =
q
4
0
r
2
+z
2
2rzcos
R
0
z
r
2
2R
2
0
r cos
z
+
R
4
0
z
2
+
R
0
z
r
2
+
2R
2
0
r cos
z
+
R
4
0
z
2
r
2
+z
2
+2rzcos
and we know that the total charge distribution induced on the plane is given by
p
=
0
1
r
dV
d
39
and so we nd that the derivative of the potential with respect to is given as
dV
d
= q
rR
3
0
sin
z
2
r
2
2R
2
0
r cos
z
+
R
4
0
z
2
3/2
+
rR
3
0
sin
z
2
r
2
+
2R
2
0
r cos
z
+
R
4
0
z
2
3/2
rzsin
(r
2
+z
2
2rzcos)
3/2
rzsin
(r
2
+z
2
2rzcos)
3/2
thus we know that
p
=
q
4
R
3
0
sin
z
2
r
2
2R
2
0
r cos
z
+
R
4
0
z
2
3/2
+
R
3
0
sin
z
2
r
2
+
2R
2
0
r cos
z
+
R
4
0
z
2
3/2
zsin
(r
2
+z
2
2rzcos)
3/2
q
4
zsin
(r
2
+z
2
2rzcos)
3/2
and since we are integrating over the plane this means =

2
and da = r sinddr and
q
p
=
Z
space
da
which will give us
q
p
=
q
4
Z
2
0
d
Z

R
0
r
R
3
0
z
2
r
2
+
R
4
0
z
2
3/2
+
R
3
0
z
2
r
2
+
R
4
0
z
2
3/2
z
(r
2
+z
2
)
3/2
z
(r
2
+z
2
)
3/2
which becomes
q
p
=
q
2
Z

R
0
R
3
0
rdr
z
2
r
2
+
R
4
0
z
2
3/2
+
q
2
Z

R
0
zrdr
(r
2
+z
2
)
3/2
solving this integral using u substitution yields
q
p
=q
z
2
(2)
1
R
2
0
+z
2
R
3
0
(2)
2z
2
1
R
2
0
+
R
4
0
z
2
simpyfying this term yields

q
p
=q
R
2
0
+z
2
R
2
0
z
2
z
z
2
+R
2
0
=q
z
2
+R
2
0
z
z
2
+R
2
0
and knowing that
q
p
+q
h
=q
q
h
q
= 1
q
p
q
thus
q
h
q
= 1
z
2
+R
2
0
z
z
2
+R
2
0
40
Chapter 3
Special Techniques
3.1 Laplaces Equation
The primary task of electrostatics is to nd the electric eld of a given stationary charge distribution. In
principle, this purpose is accomplished by Coulombs law, in the form of Equation 2.3
E(r) =
1
4
0
Z
r
r
2
(r
)d
Unfortunately, integrals of this type can be difcult to calculate for any but the simplest charge congura-
tions. We can attempt to solve for the potential using
V(r) =
1
4
0
Z
1
r
(r
)d
(3.1)
Still, even this integral is often too tough to handle analytically. In such cases it is fruitful to recast the
problem in differential, using Poissons equation
2
V =
1
which, together with appropriate boundary conditions, is equivalent to Equation 3.1. Very often in fact, we
are interested in nding the potential in a region where = 0. There may be plenty of charge elsewhere,
but were conning our attention to places where there is no charge. In this case Poissons equation reduces
to Laplaces equation
2
V = 0
or, written in cartesian coordinates
2
V
x
2
+

2
V
y
2
+

2
V
z
2
= 0
this formula is so fundamental to the subject that one might almost say electrostatics is the study of
Laplaces equation. At the same time, it is a ubiquitous equation, appearing in such diverse branches of
physics as gravitation and magnetism, the theory of heat, and the study of soap bubbles.
3.1.1 Boundary Conditions and Uniqueness Theorems
Laplaces equation does not by itself determine V; in addition, a suitable set of boundary conditions must
be supplied. The proof that a proposedset of boundary conditions will sufce is usually presented in the
41
form of a uniqueness theorem. There are many such theorems for electrostatics, all sharing the same
basic format-Ill show you the two most useful ones
First uniqueness theorem: The solution to Laplaces equation in some volume V is uniquely deter-
mined if V is specied on the boundary surface S.
Proof: Imagine a region and its boundary, there could be islands inside, so long as V is given on all
their surfaces; also the outer boundary could be at innity, where V is ordinarily taken to be zero. Suppose
there were two solutions to Laplaces equation
2
V
1
= 0
2
V
2
= 0
both of which assume the specied value on the surface. I want to prove that they must be equal. The trick
is to look at the difference
V
3
=V
1
V
2
This obeys Laplaces equation
2
V
3
=
2
V
1
2
V
2
and it takes the value zero on all the boundaries (since V
1
and V
2
are equal there). But Laplaces equation
allows no maxima or minima-all extrema occur on the boundaries. So the maximum and minimum of V
3
are both zero. Therefore V
3
must be zero, and hence
V
1
=V
2
The uniqueness theorem is a license to your imagination. It doesnt matter how you come to your solution;
if a) it satises Laplaces equation and (b) it has the correct values on the boundaries, then its right.
3.1.2 Conductors and the Second Uniqueness Theorem
The simplest way to set boundaries for an electrostatic problem is to specify the value of V on all the
surfaces surrounding the region of interest. And this situation often occurs in practice. In the laboratory,
we have conductors connected to batteries, which maintain a given potebtial, or groundm which is the
experimentalists word for V = 0.
Second uniqueness theorem: In a volume V surrounded by conductors and containing a specied
charge density , the electric eld is uniquely determined if the total charge on each conductor is given.
(The region as a whole can be bounded by another conductor, or else unbounded.)
3.2 Multipole Expansion
3.2.1 Approximate Potentials at Large Distances
If you are very far away froma localized charge distribution, it looks like a point charge, and the potential
is, to good approximation, (1/4
0
)Q/r, where Q is the total charge. We can expand the potential as
V(r) =
1
4
0
n=0
1
r
(n+1)
Z
(r
)
n
P
n
(cos
)(r
)d
(3.2)
or more explicitely
V(r) =
1
4
0
1
r
Z
(r
)d
+
1
r
2
Z
r
cos
(r
)d
+
1
r
3
Z
(r
)
2
3
2
cos
2
1
2
(r
)d
+...
42
this is the desired result- the multipole expansion of V in powers of 1/r. The rst term n = 0 is the
monopole contribution. The second (n = 1) is the dipole term; the third is the quadrupole term; the
fourth is the octopole term and so on. As it stands Equation 3.2 is exact, but it is useful primarily as an
approximation scheme: The lowest nonzero term in the expansion provides the approximate potential at
large r, and succesive terms tell us how to improve the approximation if greater precision is required.
3.2.2 The Monopole and Dipole Terms
3.3 Problems and Solutions
Problem # 1
Consider a cube made of 6 conducting plates of size aa , that encloses the space a/2 x, y, z a/2.
The plates at z =a/2 are held at potentials V
0
, respectively, while the other four sides are all grounded
(V = 0).
(a). Solve the boundary value problem for V(x, y, z) inside the cube.
First we must solve the Laplace equation
2
V =

2
V
x
2
+

2
V
y
2
+

2
V
z
2
= 0
As always, we look for solutions that are products
V(x, y, z) = X(x)Y(y)Z(z)
Plugging this into the Laplace equation we nd
1
X
d
2
X
dx
2
+
1
Y
d
2
Y
dy
2
+
1
Z
d
2
Z
dz
2
= 0
thus we know that
1
X
d
2
X
dx
2
= c
1
1
Y
d
2
Y
dy
2
= c
2
1
Z
d
2
Z
dz
2
= c
3
which implies that
c
1
+c
2
+c
3
= 0
we can now solve this second order differential equation. One of these coefcients must be negative,
this can only be c
3
because we know that this must be an oscillatory solution. In this case we nd
c
3
=(c
1
+c
2
) c
1
= l
2
c
2
= k
2
c
3
=(l
2
+k
2
)
and so
d
2
X
dx
2
= l
2
X
d
2
Y
dy
2
= k
2
Y
d
2
Z
dz
2
=(l
2
+k
2
)Z
43
the solutions to these differential equations are
X(x) = Asin(lx) +Bcos(lx)
Y(y) = Csin(ky) +Dsin(ky)
Z(z) = Ee
l
2
+k
2
z
+Fe
l
2
+k
2
z
Boundary conditions
our boundary conditions are given as
(i) V = 0 when y =a/2
(ii) V = 0 when x =a/2
(iii) V = V
0
when z =a/2
from (i) we can see that
Y(a/2) =Csin
l
a
2
+Dcos
l
a
2
= 0 Y(a/2) =Csin
l
a
2
+Dcos
l
a
2
= 0
and from (ii) we can see that
X(a/2) = Asin
k
a
2
+Bcos
k
a
2
= 0 X(a/2) =Asin
k
a
2
+Bcos
k
a
2
= 0
from the rst and second boundary condition we nd
C = 0 A = 0
and also
l =
n
a
k =
m
a
l
2
=
n
2
2
a
2
k
2
=
m
2
2
a
2
where n and m are positive integers. Since the given boundary conditions imply an odd potential we
can see that
Z(z) =Z(z) =
l
2
+k
2
z
+Fe
l
2
+k
2
z
Ee
l
2
+k
2
z
+Fe
l
2
+k
2
z
and so we nd that
F =E
and this now gives
Z(z) = E
l
2
+k
2
z
e
l
2
+k
2
z
= E2sinh(
l
2
+k
2
z)
combining the remaining constants we are left with
V(x, y, z) =Ccos
n
a
x
cos
m
a
y
sinh(
l
2
+k
2
z)
where C has absorbed the other constants, we can also write this as
V(x, y, z) =Ccos
n
a
x
cos
m
a
y
sinh
n
2
+m
2
z
44
we can now write this as a double sum over the integers n and m
V(x, y, z) =
n=1
m=1
C
n,m
sinh
n
2
+m
2
z
cos(nx/a)cos(my/a)
and nally the last boundary conditions (iii) gives us
V(x, y, a/2) =
n=1
m=1
C
n,m
sinh
n
2
+m
2
cos(nx/a)cos(my/a) =V
0
(3.3)
to determine the constant C
n,m
we multiply both sides by cos(n
x/a)cos(m
y/a) where n
and m
are
positive integers and integrate, this is known as Fouriers trick. For the left hand side we nd
m
sinh
n
2
+m
2
C
n,m
Z
a/2
a/2
cos
n
a
x
cos
a
x
dx
Z
a/2
a/2
cos
m
a
y
cos
a
y
dy
we also know from Fouriers trick and orthogonality that n = n
m = m
and so we get
sinh
n
2
+m
2
C
n,m
Z
a/2
a/2
cos
2
n
a
x
dx
Z
a/2
a/2
cos
2
m
a
y
dy =C
n,m
sinh
n
2
+m
2
a
2
4
and now for the right hand side of equation 1 we nd
V
0
Z
a/2
a/2
cos
a
x
dx
Z
a/2
a/2
cos
a
y
dy =V
0
4a
2
nm
2
sin
n
a
sin
m
a
and nally we can put these two solutions together to get

C
n,m
sinh
n
2
+m
2
a
2
4
=V
0
4a
2
nm
2
sin
n
a
sin
m
a
this yields
C
n,m
=
16V
0
nm
2
1
sinh
n
2
+m
2
for n and m odd
if n and m are even then C
n,m
is zero. We can now write the potential as
V(x, y, z) =
16V
0
2

n,m=1,3,5
1
nm
cos
nx
a
cos
m
a
y
sinh
n
2
+m
2
z
sinh
n
2
+m
2
(b). The potential is obviously zero at the center of the cube. What is the electric eld there?
To nd the electric eld we use
E =V =
V
x
+
V
y
+
V
z
45
and since we are evaluating the electric eld at the origin we know that
V
x
=
V
y
= 0
this leaves us with
E(x, y, z) =
V
z
=
16aV
0

n,m=1,3,5
1
nm
cos
nx
a
cos
m
a
y
cosh
n
2
+m
2
z
sinh
n
2
+m
2
n
2
+m
2
z
and so the electric eld at the origin is given by
E(0, 0, 0) =
16V
0
a

n,m=1,3,5
1
nm
n
2
+m
2
sinh
n
2
+m
2
z
Problem # 2
(a). Suppose that on the surface of a sphere of radius R there is a surface charge density () =
0
cos
2
, where
0
is a constant. What is V(r, ) inside and outside the sphere? What is the electric eld
E(r, ) inside and outside the sphere? Conrm that the eld is discontinuous at the surface of the sphere
in accord with the given charge distribution.
From equation 3.65 in Grifths we know that
V(r, ) =
l=0
A
l
r
l
+
B
l
r
l+1
P
l
(cos) (3.4)
this is the general form of a potential in which we assumed azimuthal symmetry in spherical coordi-
nates, this was derived in Grifths on page 137-139. Looking at the inside of the sphere when r < R we
nd that B
l
= 0 for all l otherwise the potential will diverge. Looking at the potential outside the sphere
we can see that A
l
=0 otherwise the potential would not satisfy the boundery condition that V 0 r .
From this we can see that
V(r < R, ) =
l=0
A
l
r
l
P
l
(cos)
V(r R, ) =
l=0
B
l
r
l+1
P
l
(cos)
we know that the potentials are the same at r = R and continous so we can set these expression equal
to each other
l=0
A
l
R
l
P
l
(cos) =
l=0
B
l
R
l+1
P
l
(cos)
and we can see that
B
l
= A
l
R
2l+1
since there is a discontinuity in the radial derivative we can use equation 2.36
V
out
r

V
in
r
r=R
=
1
0
()
46
from this we nd
l=0
(l +1)
B
l
R
l+2
P
l
(cos)
l=0
lA
l
R
l1
P
l
(cos) =
1
substituting the above expression for B

l
and some algebra yields
l=0
(2l +1)A
l
R
l1
P
l
(cos) =
1
0
cos
2
we can use our knowledge of Legendre polynomials to solve this problem

cos
2
=
P
2
(cos) +
1
2
2
3
=
P
2
(cos) +
P
0
(cos)
2
2
3
looking at only the l = 0, 2 terms in the sum and using this on the above expression yields
A
0
R
1
P
0
(cos) +5A
2
RP
2
(cos) =
1
3
0
P
0
(cos) +
2
3
0
P
2
(cos)
this allows us to solve for A
0
and A
2
as
A
0
=

0
R
3
0
A
2
=
2
0
15
0
R
1
so the potentials for inside and outside the spheres are given as
V(r < R, ) =

0
R
3
0
+
2
0
r
2
15
0
R
P
2
(cos)
V(r R, ) =

0
R
2
3
0
r
+
2
0
R
4
15
0
r
3
P
2
(cos)
We can nd the electric eld by
E =V =
V
r
r +
1
r
V
for inside the sphere we nd
E
in
=V
in
=
4
15
r
0
R
0
3cos
2
1
2
r +
r
0
3sin(2)
5R
0
knowing that the derivative of the term is

d
d
3
2
cos
2
=
3
2
d
d
1
2
[1+cos2] =
3
2
sin2
which can be simplied as
E
in
=V
in
=
2
15
r
0
R
0
3
2
sin(2)
[3cos
2
1] r
47
for the outside we nd
E
out
=V
out
=
0
3
0
R
2
r
2

1
5
R
4
0
r
4
0
(3cos
2
1)
r +
1
5
R
4
0
r
4
0
(sin2)
to showthat these elds are discontinous at the surface, one would just have to showthat at r =R

E
in
=
E
out
and this can be seen from the above expressions.
(b). Now repeat the above exercise for the surface charge density () =
0
sin
2
. Note that in this
case
V(r, ) =
2R
0
3
0

2R
0
15
0
r
R
2
P
2
(cos) r < R
2R
0
3
0

2R
0
15
0
R
r
3
P
2
(cos) r R
For this problem we can just start with
l=0
(2l +1)A
l
R
l1
P
l
(cos) =

0
sin
2
0
=

0
0
(1cos
2
)
and only looking at the l = 0, 2 case in the sum we nd
A
0
R
1
P
o
(cos) +5A
2
RP
2
(cos) =

0
P
0
(cos)
2
3
P
2
(cos)
1
3
P
0
(cos)
setting like terms equal to each other gives

A
0
=
2
0
R
3
0
A
2
=
2
0
15R
0
and so nally we nd that the potentials are given by
V(r < R, ) =
2R
0
3
0
2R
0
15
0
r
R
2
P
2
(cos)
and also
V(r R, ) =
2R
2
0
3
0
r

2R
0
15
0
R
r
3
P
2
(cos)
and the electric eld is given as
E
in
=V
in
=
2r
0
R
15
0
(3cos
2
1) r +
1
5
0
r
2
0
(sin2)
and for the outside we have
E
out
=V
out
=
2R
2
0
3
0
r
2
+
6
0
R
4
15
0
r
4
(cos
2
1)
r +
1
5
0
R
4
0
r
4
sin2
48
Problem # 3
Solutions to Laplaces equation in two dimensions using complex functions: One sometimes encoun-
ters a boundary-value problem that is essentially two-dimensional, in that, by the nature of the problem,
we can assume the electrostatic potential is constant along one Cartesian direction. In this case, it may be
possible to nd the solution to our boundary-value problem among the many solutions obtained by anal-
ysis of analytic functions of a complex variable. For this, lets consider the complex number = x +iy,
where x and y are the remaining two Cartesian coordinates for which we wish to determine V(x, y). We let
F() be an analytic function of - meaning that it is a well-behaved function with well dened derivatives.
(a). Show that F() satises Laplaces equation, i.e. that
2
F =

2
F
x
2
+

2
F
y
2
= 0
To show this we begin with
F() = F(x +iy) = x +iy
we also know that
F
x
=
F
z
z
x
=
F
z
1
2
F
x
2
=

x
F
z
=

z
F
x
=

2
F
z
2
and now for y we nd
F
y
=
F
z
z
y
=
F
z
i
2
F
y
2
= i

y
F
z
=

z
F
y
= i
2
2
F
z
2
=
2
F
z
2
because we know that the differential operator is commutative and we can say that
y
=

y
x
so we have shown that all nice well behaved complex functions satises Laplaces equation.
2
F =

2
F
x
2
+

2
F
y
2
=

2
F
z
2

2
F
z
2
= 0
Given this fact, we have indeed found two solutions to Laplaces equation in two dimensions. Express-
ing F = g(x, y) +ih(x, y), then the real-valued functions g and h are both such solutions.
Deriving from a single complex analytic function, the functions g(x, y) and h(x, y), known as conjugate
functions, possess an important relation to one another. By considering rst derivatives of F() one can
demonstrate the validity of the Cauchy-Riemann relations:
g
x
=
h
y
g
y
=
h
x
49
(b). Show that the electric eld obtained by setting V(x, y) = g(x, y) is everywhere orthogonal to that
obtained for V(x, y) =h(x, y). Based on this observation, one can state that the curves h(x, y) =h
0
describe
the eld lines for the potential V(x, y) = g(x, y), and visa versa.
To show that these two solutions are orthogonal we must take the dot products of the two solutions
E(g) = V =g(x, y) =
g
x
x+
g
y
y
E(h) = V =h(x, y) =
h
x
x+
h
y
y
the dot product must be zero for these two solutions to be orthogonal.
E(g)
E(h) =
g
x
x+
g
y
y
h
x
x+
h
y
y
=
g
x
h
x
+
g
y
h
y
= 0
but since we know that the Cauchy-Reimann equations are
g
x
=
h
y
h
x
=
g
y
using this we nd
g
x
h
x
g
x
h
x
= 0
(c). One application of this method: Find the potential V(x, y) in the wedge-shaped region between
two plates (innite half-planes) that intersect at the z axis at an angle </2, and are both at the potential
V
0
. You might consider the real or imaginary parts of the complex function F() =A
p
+B and attempt to
match the given boundary conditions. Other nice solutions obtained by this method are the eld produced
at the edge of a semi- innite charged conducting plate, and the fringing eld of a parallel plate capacitor.
Since we know that we can write a complex function as
F() = F(x +iy) = Ar
p
e
ip
+B = re
i
= r[cos() +i sin()]
where is now the angle dened in the complex plane and r is the distance from the origin, thus we
nd
F() = Ar
p
[cos(p) +i sin(p)] +B = Ar
p
cos(p) +B+Ar
p
i sin(p)
since we know that both the real solution and imaginary solution satisfy the Laplace equation we can
solve for the real part and we are guranteed that this is a unique solution.
RV(r, ) = Ar
p
cos(p) +B (3.5)
Imposing boundary conditions we nd
V(r, = 0) = Ar
p
+B =V
0
V(r, = ) = Ar
p
cos(p) +B =V
0
Subtracting the rst equation from the second we nd
Ar
p
Ar
p
cos(p) = 0 p =
2n
50
now if we add the two solutions we nd
Ar
p
+Ar
p
cos(p) +2B = 2V
0
using the expression for p we get
2Ar
p
+2B = 2V
0
B =V
0
Ar
p
inserting this into equation 3 we nd
Ar
p
cos(p) +V
0
Ar
p
= Ar
p
(cos(p) 1) +V
0
we can also see that
x = r cos(p) cos(p) =
x
r
r =
x
2
+y
2
and so we nd
V(x, y) = Ax(x
2
+y
2
)
(p1)/2
A(x
2
+y
2
)
p/2
+V
0
Problem # 4
Solve Laplaces equation by separation of variables in cylindrical coordinates, assuming there is no
dependence on z (cylindrical symmetry). [ Make sure you nd all solutions to the radial equation; in
particular, your result must accomidate the case of an innite line charge, for which of course we already
know the answer.]
We can begin by writing the potential as a product
V(s, ) = S(s)()
and solving the Laplace equation in cylindrical coordinates
2
V =
1
s
s
V
s
+
1
s
2
V
2
= 0
substituting our potential into the Laplace equation and multiplying every term by s
2
yields
s
S(s)
d
ds
s
dS(s)
ds
+
1
()
d()
d
2
=
s
S(s)
s
d
2
S(s)
ds
+
dS(s)
ds
+
1
()
d()
d
2
= 0
and letting
s
S(s)
s
d
2
S(s)
ds
+
dS(s)
ds
= p
1
1
()
d()
d
2
= p
2
p
1
+ p
2
= 0
if we now let p
1
=
2
and p
2
=
2
,we chose these constants because we were motivated by knowing
the angular solution should be oscillatory. We can now write
s
2
d
2
S(s)
ds
+s
dS(s)
ds
=
2
S(s)
d()
d
2
=
2
()
we can see that the angular solutions are given by
() = A
e
i
+B
e
i
= Asin() +Bcos() 1
51
where A = A
and B = B
. Now to solve the radial differential equation, which is a non-linear homo-

geneus second order differential equation. We can write it in a more suggestive form
s
2
S
(s) +sS
(s)
2
S(s) = 0
and now we can assume that it has a polynomial form
S(s) = s
m
S
(s) = ms
m1
S
(s) = m(m1)s
m2
substituting this into our equation gives
s
2
m(m1)s
m2
+sms
m1
2
s
m
= 0
m(m1) +m
2
= 0
m =
so we nd that the radial solution is given by
S(s) =Cs
+Ds
1
we have only found the solutions where 1, what about the solutions where = 0? We can solve
this by solving the same differential equations for this special case.
d
2
()
d
2
= 0
d
ds
s
dS(s)
ds
= 0
for these solutions to hold we must set the rst derivative equal to a constant
d()
d
= c () = c+k = 0
but since we know
() =(+2)
tells us that c = 0, which gives
() = k
where k is just a constant. Now for the radial part
s
dS(s)
ds
= d S(s) = d lns + j = 0
we have now found all solutions to the potential which can now be written as
V(s, ) = k +d lns +
=1
[C
+D
][A
sin() +B
cos()]
To show that our solution accomidates an innite line charge. We know that the electric eld of an
innite line charge is given by
E =

2s
0
s
52
and the potential is
V =
Z
s
0
E ds =

2
0
lns +k
and since we know that by symmetry, and also the fact that a line does not any well dened angles we
get
=1
[C
+D
][A
sin() +B
cos()] = 0
we nd
V(s, ) = k +d lns
Problem # 5
Find the potential outside an innitely long metal pipe, of radius R, placed at right angles to an other-
wise uniform electric eld E
0
. Find the surface charge induced on the pipe.
Given the solution found from the last problem
V(s, ) = k +d lns +
=1
[C
+D
][A
sin() +B
cos()]
the surface of the cylinder is equipotential and thus we can just set the potential equal to zero. We also
know that the potential does not go to zero at large x, far from the cylinder the potential is
V E
0
x +C x = scos
We need to apply the boundary conditions
(i) V = 0 when s = R
(ii) V = E
0
scos when s R
From boundary condition (i) we nd
V(R, ) = k +d lnR+
=1
[C
+D
][A
sin() +B
cos()] = 0
and from condition (ii) we nd
V(s >> R, ) = k +d lns +
=1
[C
+D
][A
sin() +B
cos()] =E
0
scos
53
where the term D
becomes negligable for s R, from conditions (i) and (ii) we can see that
k = 0 d = 0 A
= 0
and also that = 1. This gives us
V(s, ) =Cs +Ds
1
)Bcos()
we can now absorb B into C and D we get
V(s, ) = (C
s +D
s
1
)cos()
applying the boundary conditions again gives us
(i) C
R+
D
R
= 0 D
=C
R
2
(ii) C
scos() = E
0
scos C
=E
0
this works because we can ignore the D
/s term because it is negligable.

V(s, ) =
E
0
s +
E
0
R
2
s
cos()
To nd the induced charge we can use
=
0
V
s
we know tha
V
s
= cos
E
0
E
0
R
2
s
2
=
0
cos
E
0
E
0
R
2
s
2
s = R
= 2
0
E
0
cos
Problem # 6
A spherical surface of radius R has charge uniformly distributed over its surface with a density Q/4R
2
, except for a spherical cap at the north pole, dened by the bounding cone = .
54
(a). Before we get started here, show the following recursion relation for Legendre polynomials to be
true:
dP
l+1
(x)
dx

dP
l1
(x)
dx
(2l +1)P
l
(x) = 0 (3.6)
(for l > 0). This can be obatined by considering the Rodriguez formula. (Grifths 3.62)
The Rodriguez formula (Equation 3.62 in Grifths) is given as
P
l
(x) =
1
2
l
l!
d
dx
l
(x
2
1)
l
taking the rst derivative with respect to x on P
l+1
(x) yields
d
dx
P
l+1
(x) =
1
2
l
2(l +1)!
d
dx
l
d
2
dx
2
(x
2
1)
l+1
=
1
2
l
l!
d
dx
l
[(x
2
1)
l
+2lx
2
(x
2
1)
l1
]
in the above calculation we used the following relationships
(l +1)! = (l +1)l!
1
(l 1)!
=
l
l!
taking the rst derivative with respect to x on P
l1
(x) yields
d
dx
P
l1
(x) =
2
2
l
(l 1)!
d
dx
l
(x
2
1)
l1
=
2l
2
l
l!
d
dx
l
(x
2
1)
l1
55
and nally expanding the last term gives
(2l +1)P
l
(x) = (2l +1)
1
2
l
l!
d
dx
l
(x
2
1)
l
= (2l +1)
1
2
l
l!
d
dx
l
(x
2
1)
l1
(x
2
1)
putting all this together yields
d
dx
l
(x
2
1)
l
+2lx
2
(x
2
1)
l1
(x
2
1)
l1
[(2l +1)(x
2
1)
l
= 0
d
dx
l
2lx
2
(x
2
1)
l1
(x
2
1)
l1
2lx
2
(x
2
1)
l1
+(x
2
1)
l1
= 0
0 = 0
(b). Show that the potential inside the spherical surface can be expressed as
V =
Q
2
l0
1
2l +1
[P
l+1
(cos) P
l1
(cos)
r
l
R
l+1
P
l
(cos)
where, for l = 0, we should substitute P
l1
(cos) =1.
Integrating every term in Equation 1 gives
Z
P
l
(x)dx =
1
2l +1
[P
l+1
(x) P
l1
(x)]
Knowing that the surface charge density and the general expression of the potential inside a sphere are
given by
V(r, ) =
l=0
A
l
r
l
P
l
(cos) =
Q
4R
2
motivated by example 3.9 in Grifths we know that the coefcients A
l
are given by equation 3.84
A
l
=
1
2R
l1
Z

0
()P
l
(cos)sind
we also know from Fouriers trick
Z
b
a
P
l
cossind =
Z
b
a
P
l
(x)dx =
Z
a
b
P
l
(x)dx
We nd
A
l
=
1
2R
l1
Z

0
()P
l
(cos)sind+
1
2R
l1
Z

()P
l
(cos)sind
the rst integral is just zero because the surface charge density is zero there and the second integral is
given by
A
l
=
1
2R
l1
Z

()P
l
(cos)sind =
()
2R
l1
Z
cos
1
P
l
(x)dx
where the integration limits are given by the change of variable of the integral. From the relationship
from part (a) we nd
A
l
=

2(2l +1)
[P
l+1
cosP
l1
cos] =
Q
8(2l +1)R
l+1
[P
l+1
cosP
l1
cos]
56
where P
l+1
(1) P
l1
(1) = 0 , putting this into the general expression for the potential inside a
sphere yields
V(r < R, ) =
Q
8
l=0
r
l
P
l
(cos)
(2l +1)R
l+1
[P
l+1
cosP
l1
cos]
(c). What is the potential outside the sphere?
We know that the general expression for the potential outside of a sphere is given by
V(r > R, ) =
l=0
B
l
r
l+1
P
l
(cos)
but we know that the coefcients B
l
(Equation 3.81) can be expressed as
B
l
= A
l
R
2l+1
thus
V(r > R, ) =
l=0
A
l
R
2l+1
r
l+1
P
l
(cos) =
Q
8
l=0
R
l
(2l +1)r
l+1
[P
l+1
cosP
l1
cos]
(d). Find the magnitude and the direction of the electric eld at the origin.
We know that the electric eld is given by
E =V =
V
r
Since is not well dened at the origin and we know that the electric eld points in the z direction and
we do not compute the derivative with respect to . The electric eld is given by
E
in
=
Q
8
l=1
lr
l1
P
l
cos
(2l +1)R
l+1
[P
l+1
cosP
l1
cos] r
and so we know r = z, and = 0. We can also see that the above expression tells us the l = 1 in order
for the electric eld to not blow up at the origin or go to zero.
E =
Q
8R
2
1
3
[P
2
cosP
0
cos] z
=
Q
8R
2
1
3
3
2
cos
2
1
2
z
=
Q
4R
2
1
4
[cos
2
1] z
E =
Q
16R
2
[1cos
2
] z
and the magnitude is given by
|E| =
Q
16R
2
[1cos
2
]
57
(e). Discuss the limiting forms of the potential (b,c) and the electric eld (d) as the spherical cap
becomes either (1) very small or (2) so large that the entire remaining charged area becomes a very small
cap at the south pole.
We can see from the gure that if gets very small the sphere would be almost represented as a
spherical conductor and so
1 very small
the potentials and the electric eld would behave as
V(r < R, ) V
0
V(r > R, )
1
r
E(r < R, ) 0
and if the cap becomes very large the sphere would be represented by a point charge at the south pole
of the sphere. The potentials and electric eld would behave as
V(r < R, )
1
r
V(r > R, )
1
r
E(r < R, )
1
r
2
Problem # 7
Many nuclei have electric quadrupole moments. The quadrupole moment of a given state of a given
nucleus is typically quantied as Q = Q
33
/e where Q
i j
is dened as in Grifths Problem 3.45, and the
index 3 refers to the z axis. Now, consider a nucleus centered at the origin nding itself in a cylindrically
symmetric electric eld with a gradient E
z
/z along the z axis at the position of the nucleus.
(a). Show that the inhomogeneous eld modies the energy of the quadrupole by
W =
e
4
Q
E
z
z
For this, you might want to nd a suitable form of the electrostatic potential near the position of the
nucleus. Be sure that this potential satises the Laplace equation. You should let V = 0 at the origin.
We know that energy of a continuous charge distribution can be written as
W =
Z
(r
)(r
)d
where (r
) is the externel potential energy, where the general form can be expressed as
(r) =
1
4
l=0
1
r
n+1
Z
(r
)
n
P
n
(cos
)(r
)d
we can expend this using Taylor expansion to nd

(0) = (0) +r
(0) +
1
2
i, j
r
i
r
j
j
(0)
58
where we know that the last term is the quadrupole term, which is the one we are concerned about.
Plugging this into the work function yields
W =
1
2
Z
(r
i, j
r
i
r
j
j
(0)d
and since we are only concerned with Q

33
we know that i = j = 3 , thus we nd
W =
1
2
Z
(r
)d
(r
3
)
2

2
z
2
(0)
but we know that
2
z
2
(0) =
E
z
z
putting this into the above expression yields
W =
1
2
E
z
z
Z
(r
3
)
2
(r
)d
using the relationship from problem 3.45

Q
33
=
Z
[3(r
3
)
2
(r
3
)
2
](r
)d

Q
33
2
=
Z
(r
)d
(r
3
)
2
using this expression we nd that the work is given by
W =
Q
33
4
E
z
z
since we know that
Q
33
= eQ
we nd
W =
eQ
4
E
z
z
(b) Nuclear charge distributions can be approximated by a constant charge density throughout a
spheroidal (egg-shaped) volume of semimajor axis a and semimajor axis b. Calculate the quadrupole
moment of such a nucleus assuming that the total charge is Ze. Given that the
153
Eu(Z = 63) nucleus has
a quadrupole moment of Q = 2.5 10
25
cm
2
and a mean radius R =
a+b
2
= 7 10
13
cm, determine the
fractional difference in radius (ab)/R.
59
Using the denition of the quadrupole moment
Q
i j
=
Z
[3r
i
r
j
(r
)
2
i j
](r
)d
and since we are only looking for Q

zz
we nd
Q
zz
=(r)
Z
[2z
2
x
2
y
2
]dxdydz
This gives us the quadrupole moment in cartesian coordinates, it would be easier to solve this integral
in spherical coordinates, we can do a linear transformation to do this, the equation that models our spheroid
is given by
x
2
b
2
+
y
2
b
2
+
z
2
a
2
= 1
letting
u =
x
b
v =
y
b
w =
z
a
where the derivatives are given by
dx = bdu dy = bdv dz = adw
and so we know
dxdydz = ab
2
dudvdw
we also know that the cartesian components in spherical coordiantes are given by
u = r sincos
v = r sinsin
w = r cos
60
using the above relationships we get
Q
zz
=(r)
Z Z Z
[2(wa)
2
(bv)
2
(bu)
2
]ab
2
d
which gives
(r)
Z
2
0
Z

0
Z
1
0
[2a
2
r
2
cos
2
b
2
r
2
sin
2
cos
2
b
2
r
2
sin
2
sin
2
]ab
2
r
2
sindddr
simplifying this gives
Q
zz
=(r)
Z
2
0
Z

0
Z
1
0
[2a
2
r
2
cos
2
b
2
r
2
sin
2
]ab
2
r
2
sindddr
and using Mathematica yields
Q
zz
=(r)
8
15
ab
2
(a
2
b
2
)
but we know that the charge density is given as
(r) =
q
V
=
3q
4ab
2
V
ell
=
4
3
ab
2
subsituting this expression gives
Q
zz
=
2
5
q((ab)(a+b))
solving for ab/R gives
ab
R
=
5Q
zz
2Ze(a+b)R
=
5Q
zz
4ZeR
2
since we know that
Q
zz
= Qe
we nd that the fractional difference is given by
ab
R
=
5Q
4ZR
2
0.01
Problem # 8
Consider a dipole with moment p at a distance z
0
above a grounded conducting plane that is taken
to dene the x y plane. p is taken to lie in the x z plane, oriented at an angle to the normal (z)
of the plane. Find the torque N on the dipole. What are the equilibrium values of ? [Answer: N
=p
2
sincos y/32
0
z
3
0
)]
61
We know that the torque from an electric eld on a dipole is given as
N = pE
and we also know that the energy of a dipole is given as
E
dip
=
1
4
1
r
3
[3(p r) r p]
we can solve this problem by using the image of the dipole, we know that the components of the dipole
and the image are given by
p
dip
= psin, 0, pcos
p
im
= p
sin, 0, p
cos
p
im
= p
dip
we also know that r points in the z direction, thus
p r =p z = p
cos r = 2z
0
putting this into the electric dipole equation yields
E
im
=
1
4
1
(2z
0
)
3
[3p
cos z +[p
sin xp
cos z]] =
1
32
1
z
3
0
[p
sin x+2p
cos z]
and thus we know that the torque is
N = pE =
p
2
32z
3
0
x y z
sin 0 cos
sin 0 2cos
which yields
N =
p
2
sincos
32z
3
0
y
62
The equilibrium values of are given when the torque goes to zero, i.e
=
n
2
n 0 integer
Problem # 9
Using Eq. 3.103, calculate the average electric eld of a dipole, over a spherical volume of radius R,
centered at the origin. Do the angular intervals rst. [Note : You must express r and

in terms of x, y and
z before integrating. Compare your answer with the general theorem Eq. 3.105. The discrepency here is
related to the fact that the eld of the dipole blows up at r = 0. The angular integral is zero, but th radial
integral is innite, so we really dont know what to make of the answer. To resolve this dilemma, lets say
Eq. 3.103 applies outside a tiny sphere of radius - its contribution to E
ave
is then unambiguously zero,
and the whole answer has to come from the eld inside the sphere.
We know that the average electric eld and the electric eld from the dipole can be expressed as
E
ave
=
R
space
E
dip
d
V
E
dip
=
p
4
(2cos r +sin
)
we need to change to cartesian coordiantes so that we can evaluate this integral, i.e
r = sincos x+sinsin y+cos z

thus E
dip
in cartesian coordinates gives
E
dip
=
p
4
[3sincoscos x+3sincossin y+(2cos
2
sin
2
) z]
and so we nd that the integral can be expressed as
Z
space
E
dip
d =
p
4
Z
r
0
Z
2
0
Z

0
3sin
2
coscos
r
drdd x+
p
4
Z
r
0
Z
2
0
Z

0
3sin
2
cossin
r
drdd y
+
p
4
Z
r
0
Z
2
0
Z

0
(2cos
2
sin
2
r
drdd y
using Mathematica yields (for the angular integrals)
Z
ang
E
dip
d = 0
and we know that the integral for the radial component (at the origin) yields
Z
rad
E
dip
d =
thus we nd that
E
ave
= 0
which seems very weird? There must be a delta function in the center of the sphere.
(b). What must the eld inside the sphere be, in order for the general theorem (3.105) to hold?
[Hint : since is arbitrarily small, were talking about something that is innite at r = 0 and whose
integral over an innitesimal volume is nite.][Answer : (p/3
0
)
3
(r)]
63
[Evidently, the true eld of a dipole is
E
dip
(r) =
1
4
0
1
r
3
[3(p r) rp]
1
3
0
p
3
(r)
You may well wonder how we missed the delta-function term when we calculated the eld back in
Sect. 3.4.4. The answer is that the differentiation leading to Eq. 3.103 is perfectly valid except at r = 0 ,
but we should have known (from our experience in Sect 1.5.1) that the point r = 0 is problematic.
Since we know that the average electric eld inside the sphere is not zero we can write the electric
dipole as
E
dip
=
1
4r
3
[3(p r) rp] +K
3
(r)
where K is a non-zero coefcient that we need to solve for. we can solve for K by solving the average
electric eld in terms of a delta function
E
ave
=
R
K
3
(r)d
V
=
3K
4R
3
setting this expression to equation 3.105 we nd
3K
4R
3
=
p
4
0
R
3
K =
p
3
0
thus the correct expression for the electric dipole is given by
E
dip
=
1
4r
3
[3(p r) rp]
p
3
0
3
(r)
64
Chapter 4
Electric Fields in Matter
Problem # 1 Forces on dipoles
(a). A dipole p with xed magnitude is placed in an external electric eld E(r) that varies with position.
Show that there is a force on the dipole, and nd how this force depends on the orientation of the dipole
and on the variation of the eld. [Answer: F = (p )E]
So we know that
F
+
= qE
+
F
=qE
and we also know that the total force is given by

F
tot
= F
+
+F
= q(E
+
E
) = qE
where the components of E are given by
E
x
= (E
x
) d E
y
= (E
y
) d E
z
= (E
z
) d
where d is the displacement vector. Thus we know
E = (d )E
thus the force is given by
F = (p )E
(b). Find the force and torque on a dipole in the eld of a point charge. Let the charge q be at the
origin, and the dipole p = p(sin
0
x +cos
0
z) be at the point (0, 0, z
0
). Also, nd the force on q, and
verify Newtons third law. [Answer: the force on q is
F =
p
0
q
4
0
z
3
0
(sin
0
x+2cos
0

z)
We know that the potential of a point charge can be written as (in cartesian coordinates)
V
c
=
q
4
1
x
2
+y
2
+z
2
65
and we know that the electric eld can be written as
E =V =
q
4
x
(x
2
+y
2
+z
2
)
3/2
x+
y
(x
2
+y
2
+z
2
)
3/2
y+
z
(x
2
+y
2
+z
2
)
3/2
z
now that we have the electric eld we can nd the force

F =
psin
0
x
E+ pcos
0
z
E
which yields
F(0, 0, z
0
) =
pq
4
[sin
0
x2cos
0
z]
and the torque is given by
N(0, 0, z
0
) = pE =
x y z
psin
0
0 pcos
0
0 0
q
4z
2
0
=
pqsin
0
4z
2
0
y
To verify Newtons third law we need to show the force on q from the dipole is equal and opposite of
the force on the dipole from q, thus we can start with the defenition for the electric eld of a dipole
E
dip
(r) =
1
4r
3
[3(p r) r p]
and we know that the force is given by
F = qE
dip
we also know that
r = z p = p(sin
0
x+cos
0
z) (p r) r = pcos
0
z
thus we can see that the force is
F =
pq
4z
3
0
[3cos
0
z sin
0
xcos
0
z] =
pq
4z
3
0
[2cos
0
z sin
0
x]
thus we can see that
F = qE
dip
=qE
c
We know that the equilibrium values for are given when
=
n
2
n 0
Problem # 2 Capacitors with dielectric slabs
The space between the plates of a parallel capacitor (Figure 1) is lled with two slabs of linear dialectric
material. Each slab has thickness a, so the total distance between the plates is 2a. Slab 1 has a dialectric
constant of 2, and Slab 2 has a dialectric constant of 1.5. The free charge density on the top plate is and
onthe bottom plate .
66
Figure 1
a). Find the electric displacement D in each slab.
Since we know what the free charge is on both the top plate of the capacitor and the bottom plate of
the capacitor we can apply the integral form of Gausss law to nd the displacement, using
I
top
D da = Q
f
we can nd what the displacement if we place a Gaussian pillbox encompassing the top plate of the
capacitor and Slab 1
Q
f
=
f
A
I
top
D
1
da =
f
A da = A( z)
where z points in the direction of the electric eld, which is downward. So for Slab 1 we nd the
displacement to be
D
1
=
f
( z) = ( z)
and for the displacement of Slab 2 we can place a Gaussian pillbox encompassing the bottom plate of
the capacitor and Slab 2
Q
f
=
f
A
I
bot
D
2
da =
f
A da = A( z)
thus
D
2
=
f
( z) =( z)
and the magnitudes are given by
D
1
= D
2
=
b). Find the electric eld E in each slab.
given that we know what D is in each slab we can easily nd E
E
1
=
D
1
1
= 2
0
E
1
=

2
0
z
and
E
2
=
D
2
2
=
3
2
0
E
2
=
2
3
0
z
E
1
=

2
0
E
2
=
2
3
0
67
c). Find the polarization P in each slab.
We know that the polarization is dened as
P =
0
e
E
where
e
is dened as
0
=
r
=
0
(1+
e
)
e
=
r
1
1
= 1
2
=
1
2
for Slab 1 and 2 we nd
P
1
=

2
z P
2
=
3
z
P
1
=

2
P
2
=

3
d). Find the potential difference between the plates.
we know that the potential difference is given by the following integral
V =|V
1
V
2
| =|
Z
a
0
E
1
dl
Z
2a
a
E
2
dl| =
a
2
0
+
2a
3
0
given that dl = a z we nd
V =
7a
6
0
e). Find the location and amount of all bound charge.
given that we know the polarization we can nd what the bound charge is by using
b
= P n
where n points outward with respect to the dialectric, so for Slab 1 we nd
b
=
2
Slab 1 top surface
b
=

2
Slab 1 bottom surface
and for Slab 2 we nd
b
=
3
Slab 2 top surface
b
=

3
Slab 2 bottom surface
this is illustrated in Figure 2.
68
Figure 2
f). Now that you know all the charge (free and bound), recalculate the eld in each slab, and conrm
your answer to b.
Using Figure 2 we can nd the total charge
t
for Slab 1 and Slab 2. For the top of Slab 1 we add up
all the charge in the dashed box labeled total-top(1) which yields
t
=
2
=

2
Top of Slab 1
for the bottom of Slab 1 we add up all the charge in the box labeled total-bottom(1) which yields
t
=

2
3
+

3
=
2
bottom of Slab 1
and now for the bottomof Slab 2 we add up all the charge in the dashed-dot box labeled total-bottom(2)
which yeilds
t
=

3
=
2
3
bottom of Slab 2
and nally for the top of Slab 2 we add up all the charge in the dashed-dot box labeled total-top(2)
which yields
t
=
3
+

2
2
+ =
2
3
Top of Slab 2
and the magnitude of the electric eld is given by
E
1
=

t
0
=

2
0
E
2
=

t
0
=
2
3
0
which is the same thing we found in part b.
part 2: problem 2
Suppose you have enough linear dialectric material, of dialectric constant
r
, to hal f -ll a parallel-
plate capacitor (Figure 3). By what fraction is the capacitance increased when you distribute the material
as in Fig.3 (a)? How about Fig.3 (b)?
Figure 3
For Figure 3 (a) we can see that the displacement is given by
I
D da = Q
f
= A D =
this is the displacement inside the dialectric, with this we can now nd what the electric eld is inside
the dialectric
E =
D
E
dia
=

69
we also know that the electric eld between the dialectric and the top plate of the capacitor is given by
E
air
=

0
given that the defenition of capacitance is given by
C
1/2
=
Q
V
1/2
we are now in a position to nd what the potential is between the capacitor
V
1/2
=
Z
d/2
0
E
air
dl +
Z
d
d/2
E
dia
dl =
d
2
0
+
d
2
=
d
2
0
+
1
=
d
2
(
r
+1)
thus the capacitance is given by
C
1/2
=
Q
V
1/2
=
A
V
1/2
=
2A
d(
r
+1)
and since we know that a capacitor with no dialectric has a capacitance given by
C =
A
0
d
thus the ratio of the capacitance with dialectric to capacitance without dialectric for Figure 3 (a) is
given by
C
1/2
C
=
2
0
(
r
+1)
=
2
r
(
r
+1)
From Figure 3 (b) we can see that the area has been decreased by 1/2 and the dialectric now lls this
entire 1/2-space, thus we can treat this as two separate capacitors one in which the 1/2 area is lled with
the dialectric and another 1/2 area that has no dialectric. We nd that the total capacitance is given by
C
1/2
=
A
2d
+
A
0
2d
=
A
2d
( +
0
)
and we know that the capacitance with no dialectric is given by
C =
A
0
d
thus the ratio gives
C
1/2
C
=
+
0
2
0
=

r
+1
2
Problem # 3
A spherical conductor, of radius a, carries a charge Q. It is sorrounded by linear dialectric material of
susceptibility
e
, out to radius b. Find the energy of this conguration.
We know that the total work done, as we build up the free charge from zero to its nal conuguartion
is given by
W =
1
2
Z
D Ed
70
We can nd D by using
I
D da = Q
f
D =
Q
4r
2
r r > a
we know that P=D=E=0 inside the sphere. Now that we know D we can easily nd E,
E =
D
=
Q
4r
2
r a < r < b
and
E =
Q
4
0
r
2
r r > b
now that we have D and E we can now nd what the energy of this conguration is
W =
Q
2
2
Z
b
a
1
(4)
2
r
2
drd+
Z

b
1
(4)
2
0
r
2
drd
knowing that d= 4 when integrating over all solid angles allow us to symplify this as
W =
Q
2
8
Z
b
a
1
r
2
dr +
1
0
Z

b
1
r
2
dr
=
Q
2
8
1
a
1
b
+
1
1
b
we can also write this as

W =
Q
2
8
1
a
1
b
+
r
1
b
=
Q
2
8
1
a
+(
r
1)
1
b
but since we know
r
= 1+
e
=
0
(1+
e
)
we nd
W =
Q
2
8
0
(1+
e
)
1
a
+
e
1
b
Problem # 4
The dielectric constant of air is 1.00059. From this, determine the mean polarizability of atmospheric
molecules. Compare this result to the atomic polarizabilities listed, for example, in Grifths Table 4.1.
We can write the polarization as
P =
e
0
E
The polarization in a dialectric is also dened as
P = np
where p is the dipole moment and n is the particle density. We also know that the dipole moment can
be dened as
p =E
other
where is the polarizability and E
other
is the total electric eld, the applied eld and the enduced eld
E
other
= EE
sel f
71
we need to nd out what the self induced eld is. The external eld produced by the sphere itself is a
pure dipole. The total eld inside the conducting sphere is 0, because a conductor is an equipotential, thus
the induced eld inside must be E
0
, this eld is needed to cancel the applied uniform eld, so we nd
E
sel f
=E
0
If we treat the sphere as a conducting sphere we nd that the dipole moment goes as
p = 4
0
a
3
E
0
where a is the radius, thus E
sel f
can be written as
E
sel f
=
p
4
0
a
3
substituting this in E
other
we nd
E
other
= E+
p
4
0
a
3
we also know that
E
sel f
=
np
3
0
=
P
3
0
thus we nd
P = np = n
E+
P
3
0
there is a linear relationship between P and E, and the susceptibility which is given by
e
=
n/
0
1n/(3
0
)
if we solve for the atomic polarizability we nd
=

0
n
e
1+
e
/3
=
3
0
n
1
+2
and usin the ideal gas law we nd
n =
N
V
=
kT
P
we nd for /4
0
4
0
=
3kT
4P
1
+2
1.9010
30
1
m
3
using T = 293 K and 110
5
Pa for the pressure, along with the dialectric constant given.
Problem # 5
A hollow dielectric sphere, with dielectric constant /
0
= , inner radius a and outer radius b, is
placed in a uniform applied electric eld E
0
z. The presence of the sphere changes the eld. Find the eld
in the three dened regions, i.e. r < a, a < r < b and r > b. What is the eld at the center of the spherical
shell? What is the dipole moment of the dielectric medium? [for the last part, youll nd:
p
z
=
4
0
(a
3
b
3
)b
3
E
0
(1+2)(+1)
2a
3
(1)
2
b
3
(2+)(1+2)
72
Using the general solution for the potential in spherical coordinates we nd thet the potentials are given
by
V
1
=

l=0
A
l
r
l
P
l
cos r < a
V
2
=

l=0
B
l
r
l
+
C
l
r
l+1
P
l
cos a < r < b
V
3
=

l=0
D
l
r
l+1
P
l
cosE
0
r cos
and the boundary conditions are given by
(i) V
1
= V
2
r = a
(ii) V
2
= V
3
r = b
(iii)
V
2
r
=
0
V
1
r
r = a
(iv)
V
2
r
=
0
V
3
r
r = b
thus we have four unknown constants and four boundary conditions. Lets gure out what these con-
stants are using the boundary conditions. From 1 we nd
l=0
A
l
a
l
P
l
cos =
l=0
B
l
a
l
+
C
l
a
l+1
P
l
cos
which simplies to
A
l
a
l
= B
l
a
l
+
C
l
a
l+1
l = 1 A = B+
C
a
3
l = 1
if
l = 1 A
l
= B
l
=C
l
= 0
from this and from the next boundary condition we know that l = 1 for all cases. From the second
boundary condition we nd (using l = 1)
Bb
3
+C = DE
0
b
3
73
from boundary condition number 3 we nd
0
A =
B
2C
a
3
0
which simplies to
A = B
2C
a
3
and from our last boundary condition we nd
2D
b
3
E
0
= B
2C
b
3
so now we have four equation and four unknowns, we can do this by brute force or just use use
Mathematica. The four equations are summarized as
A = B+
C
a
3
Bb
3
+C = DE
0
b
3
A = B
2C
a
3
2D
b
3
E
0
= B
2C
b
3
Mathematica yields
A =
9b
3
E
0
2a
3
(1)
2
b
3
(2+)(1+2)
B =
3b
3
E
0
(1+2)
2a
3
(1)
2
b
3
(2+)(1+2)
C =
3a
3
b
3
E
0
(1)
2a
3
(1)
2
b
3
(2+)(1+2)
D =
(a
3
b
3
)b
3
E
0
(1+2)(1)
2a
3
(1)
2
b
3
(2+)(1+2)
plugging these constants into the general expression for all potentials yield
V
1
=
9b
3
E
0
2a
3
(1)
2
b
3
(2+)(1+2)
r cos =
9b
3
E
0
2a
3
(1)
2
b
3
(2+)(1+2)
z r cos
and for the other two we nd
V
2
=
3b
3
E
0
(1+2)
2a
3
(1)
2
b
3
(2+)(1+2)
z +
3a
3
b
3
E
0
(1)
2a
3
(1)
2
b
3
(2+)(1+2)
z
(x
2
+y
2
+z
2
)
3/2
and the nal potential is given by
V
3
=
(a
3
b
3
)b
3
E
0
(1+2)(1)
2a
3
(1)
2
b
3
(2+)(1+2)
z
(x
2
+y
2
+z
2
)
3/2
E
0
z
74
The electric eld in all three regions is given by
E =V
so
E
1
=
9b
3
E
0
2a
3
(1)
2
b
3
(2+)(1+2)
z r < a
the electric eld inside the dialectric is given by
E
2
=
3b
3
E
0
(1+2)
2a
3
(1)
2
b
3
(2+)(1+2)
z +
3a
3
b
3
E
0
(1)
2a
3
(1)
2
b
3
(2+)(1+2)
3xz
(x
2
+y
2
+z
2
)
5/2
x
3yz
(x
2
+y
2
+z
2
)
5/2
y
3z
2
(x
2
+y
2
+z
2
)
5/2
z +
1
(x
2
+y
2
+z
2
)
3/2
z
and nally for the electric eld on the outside we nd, using
K =
(a
3
b
3
)b
3
E
0
(1+2)(1)
2a
3
(1)
2
b
3
(2+)(1+2)
E
3
= K
3xz
(x
2
+y
2
+z
2
)
5/2
x
3yz
(x
2
+y
2
+z
2
)
5/2
y
3z
2
(x
2
+y
2
+z
2
)
5/2
z +
1
(x
2
+y
2
+z
2
)
3/2
z
E
0
z
The eld in the center of the dialectric is given by
E
center
=
9b
3
E
0
2a
3
(1)
2
b
3
(2+)(1+2)
z
We know that the potential in terms of the dipole moment is given by
V =
p r
4
0
r
2
=
p
z
cos
4
0
r
2
and looking at the potential on the outside we can see that the potential from the dipole moment is just
the rst term in the potential V
3
, the E
0
is from the applied eld, we are only interested in the potential
from the dipole, i.e
p
z
=
4
0
(a
3
b
3
)b
3
E
0
(1+2)(1)
2a
3
(1)
2
b
3
(2+)(1+2)
Problem # 6
Consider two insulating media with dielectric constants
1
/
0
=
1
and
2
/
0
=
2
, placed together
with a planar interface between them. In the region of ?1 there is an innite line charge with linear charge
density
1
, parallel to the interface at perpendicular distance
75
a. Use the method of images to nd the electric eld in both media.
We can see from diagram (2) that the potential in the top region is given by
V
t
=

2
1
ln|z a| +

2
1
ln|z +a|
and the potential at the bottom is given by
V
b
=

2
2
ln|z a|
we also know that the boundary conditions are given by
V
t
=V
b
z = 0
from continuity. Our othert boundary condition is given as
1
V
t
z
=
2
V
b
z
From boundary condition number 1 we can see that
+
=

1
and from condition 2 we nd
plugging in
from the last equation into the previous one yields

+
=

1
2
(
)
thus
2
+
1
plugging this into either equation yields
1+

2
2
+
1
2
2
2
+
1
76
plugging these solutions into our potentials we nd
V
t
=

2
1
ln|z a| +
2
+
1
1
2
1
ln|z +a|
and the electric eld at the top is given by
E
t
=

2
1
|z a|
2
+
1
1
2
1
1
|z +a|
and the potential at the bottom is given by
V
b
=
1
2
2
2
2
2
+
1
ln|z a|
and the electric eld is given as
E
b
=
1
2
+
1
1
|z a|
Show that the force per unit length on the line charge is
dF
dL
=

2
(
1
2
)
4
1
a(
1
2
)
We know that
F = qE dF = dqE
but
dq = dl
and since we know that there is no self induced force on we must use the
term as the one con-

tributing to the force, so we nd
dF
dL
=E
1
=
2
(
1
2
)
4
1
(
1
+
2
)
Note that the sign of the force depends on the difference
1
2
. Explain physically whats going on
here.
We can see from the expression for the force that if
1
is greater than
2
the force will be pointing in
the z direction and if we have the inverse it will point in the - z direction.
Problem # 7
77
Two long cylindrical sheets of metal (radii r
1
and r
2
with r
2
> r
1
) are arranged coaxially. The plates
connected to a battery that maintains a potential difference V between the sheets. The region between the
conductors is lled with a material of conductivity and permittivity .
a) Determine the capacitance per unit length of this system.
We know that the capacitance and the potential are given by
C =
Q
V
V =
Z
E dl
we can nd the electric eld by using Gausss law, i.e
Z
E da =
Q
da = 2rL r
we nd the electric eld to be given as
E =
Q
2rL
r
knowing the electric eld allows us to nd the potential
V =
Z
r
1
r
2
Q
2rL
dr =
Q
2L
ln
r
2
r
1
knowing this allows us to nd the capacitance per unit length

C
L
=
2
ln
r
2
r
1
b) Use Ohms law to calculate the electric current per unit length between the conducting shells.
We know that Ohms law can be expressed as
I =
Z
J da =
Z
E da
using Gausss law we nd
Z
E da =
Q
Q = L
thus we nd that the current per unit length is given by
I
L
=

=
2
ln(r
2
/r
1
)
V
c) Suppose the battery that maintains the potential difference V is suddenly disconnected from the
circuit. Show that charge will leak off the two plates of this capacitor as an exponential function of time.
What is the exponential time constant? You should obtain a simple function of and .
We know that once the battery is disconnected the current will be given as
I =
dQ
dt
=
Q
78
and this is a simple integral of the form
Z
Q
f
Q
i
dQ
Q
=
Z
t
0
dt
which yields
Q
f
= Q
i
e
t
and the exponential time constant is given by
k =

d) Show that the total energy dissipated by Joule heating as the capacitor discharges completely equals
the electrostatic energy that was originally stored in the capacitor.
We know that the work can be expressed as
W =
Z
Pdt
we know that the power given for Joule heating can be expressed as
P = I
2
R
thus we nd
W =
Z
I
2
Rdt
but we also know that
I =
dQ
dt
=Q
i
t
I = Q
i
t
we also know that
R =
V
I
=
ln
r
2
r
1
2L
thus we nd that the work is given by
W =
Q
i
2
ln
r
2
r
1
1
2L
Z

0
e
2
t
dt =

2
Q
i
2
ln
r
2
r
1
1
2L
this simplies into
W =
Q
2
4L
ln
r
2
r
1
=
1
2
Q
2L
ln
r
2
r
1
Q =
1
2
QV
we know that the work from a capacitor is given by
W =
1
2
CV
2
=
1
2
QV
thus we have just shown that these two expressions are equivalent.
79
Chapter 5
Magnetostatics
5.1 The Lorentz Force Law
The basic problem in classical electrodynamics we have a collection of charges q
1
, q
2
, q
3
... which are
known as the source charges, and we want to know the force they exert on a charge Q. Up till now we
were dealing with charges that were at rest electrostatics and the time has come to consider the forces
between charges in motion. Imagine that we have two wires seperated by a few centimeters and when I
turn on a current so that it passes up one wire and down the other and the wires jump apart. How do you
explain this? Whatever force accounts for the attraction of parallel currents and repulsion of antiparallel
ones is not electrostatic in nature. The force responsible for the observed phenomenom is the magnetic
force, we can understand this using Figure 1
From Figure 1 we can see that a cross product will allow us to solve for the force. The magnetic force
in a charge Q , moving with a velocity v in a magnetic eld B, is
F
mag
= Q(vB) (5.1)
This is known as the Lorentz force law. In the presence of both electric and magnetic elds, the net
force on Q would be
F = Q[E +(vB)] (5.2)
If we take a look at the Lorentz force law we will nd some bizarre particle trajectories.
80
5.1.1 Cyclotron Motion
A charged particle moving about a magnetic eld in circular motion, with the magnetic eld providing
the centripital acceleration is represented in the gure above. A uniform magnetic eld points into the
page The charge Q is moving counterclockwise, the magnetic force points inward and has a magnitude
QvB. We can write
F
cent
= ma
cent
= m
v
2
R
in order to sustain circular motion
F
cent
= F
mag
= QvB = m
v
2
R
which can also be written
QvB = m
v
2
R
or p = QBR (5.3)
Equation 5.3 is known as the cyclotron formula because it describes the motion of a particle in a
cyclotron. We have also assumed that the particle moves in a plane perpendicular to B. If it starts out with
some additional speed v parallel to B, this component of the motion is unaffected by the magnetic eld,
and the particle moves in a helix.
Another feature of the magnetic force law is
Magnetic forces do no work
For if Q moves an amount dl = vdt , the work done is
dW
mag
= F
mag
dl = Q(vB) vdt = 0
this follows because (vB) is perpendicular to v, so (vB) v = 0.
5.2 Currents
The current in a wire is the charge per unit time passing a given point. Current is measured in coulombs-
per-second, or amperes (A)
1 A = 1 C/s
81
A line charge traveling down a wire at speed v constitutes a current
I = v
Current is actually a vector
I = v
The magnetic force on a segment of current-carrying wire is given by
F
mag
=
Z
(vB)dq =
Z
(vB)dl =
Z
(I B)dl
if I and dl both point in the same direction, we can write this as
F
mag
=
Z
I(dl B) = I
Z
(dl B) (5.4)
5.2.1 Surface Currents
When charge ows over a surface, we describe it by the surface current density, K, dened by consider-
ing a ribbon of innitesimal width dl , running parallel to the ow. If the current in this ribbon is dI,
the surface current density is
K
dI
dl
this is simply the current per unit widthperpendicular to f low. This can also be written with a
surface charge density
K = v
and the magnetic force on the surface current is
F
mag
=
Z
(vB)da =
Z
(KB)da (5.5)
5.2.2 Volume Currents
When charge ows thru a three-dimensional region, we describe it by the volume current density, J,
dened by considering a tube of innitesimal cross section da , running parallel to the ow. If the
current in this ribbon is dI, the volume current density is
J
dI
da
(5.6)
this is simply the current per unit area perpendicular to f low. This can also be written with a
surface charge density
J =v
and the magnetic force on the surface current is
F
mag
=
Z
(vB)d =
Z
(JB)d (5.7)
Equation 5 can be written as
I =
Z
S
Jda =
Z
S
J da
82
which describes the current crossing a surface S. The dot product serves neatly to pick out the appror-
iate component of da. In particular, the total charge per unit time leaving a volume V is
I
S
J da =
Z
V
( J)d
because charge is conserved, whatever ows out thru a surface must come at the expense of that
remaining inside, this can be mathematically stated as
J =
t
(5.8)
this is known as the continuity equation. For future reference, lets summarize what we know
n
i=1
( )q
i
v
i
Z
line
( )Idl
Z
sur f ace
( )Kda
Z
volume
( )Jd
this corresponds to q dl da d for the various charge distributions.
5.3 The Biot-Savart Law
5.3.1 The Magnetic Field of a Steady Current
The magnetic eld of a steady current ius given by the Biot-Savart Law
B(r) =

0
4
Z
I

r
r
2
=

0
4
I
Z
dl
r
2
The integration is along the current path, in the direction of the ow; dl
is an element of length along

th wire, and r
, as always, is the vector from the source to the point r. The constant
0
is called the
permeability of free space.
0
= 410
7
N/A
2
These units are such that B itself comes out in newtons per ampere-meter (as required by the Lorentz
force law) or teslas (T)
1 T = 1 N/(A m)
The Biot-Savart law plays a role analogous to Coulombs law in electrostatics. The inverse square law
is common in both laws.
83
5.3.1.1 Wire segment
The magnetic eld a distance s from a long straight wire carrying a steady current I is given by
B =

0
I
4s
(sin
2
sin
1
)
this is the general solution for any line segment, thus the magnetic eld of an innitely long wire would
be given as.
B =

0
I
2s
(5.9)
because
2
=
1
=/2.
5.3.1.2 Circular loop of wire
The magnetic elda distance z above the center of a circular loop of wire with radius R, which carries a
steady current I is given as
B(z) =

0
I
2
R
2
(R
2
+z
2
)
3/2
For surface and volume currents the Biot-Savart law becomes
B(r) =

0
4
Z
K(r
r
2
da
and B(r) =

0
4
Z
J(r
r
2
d
5.4 The Divergence and Curl of B

5.4.1 Straight-Line Currents
The magnetic eld of an innitely straight wire is shown as
According to Equation 7, the integral of B around a circular path of radius s, centered at the wire is
I
B dl =
I

0
I
2s
dl =

0
I
2s
I
dl =
0
I
84
Notice that the solution is independent of s, thats because B decreases at the same rate that the circum-
ference increases. Now suppose we have a bundle of straight wires. Each wire that passes thru our loop
contributes
0
I, and those outside contribute nothing. The line integral will then be
Z
B dl =
0
I
enc
(5.10)
where I
enc
stands for the total current enclosed by the integration path. If the ow of charge is
represented by a volume current density J, the enclosed current is
I
enc
=
Z
J da
with the integral taken over the surface bounded by the loop. Applying Stokes theorem to Equation
5.8 gives
Z
(B) da =
0
Z
J da
and hence
B =
0
J (5.11)
We also know that the divergence of B is given by
B = 0
the divergence of the magnetic eld is 0.
5.4.2 Applications of Amperes Law
The equation for the curl of B
B =
0
J
is called Amperes law (in differential form). It can be converted to integral form by the usual device
of applying one of the fundemental theorems-in this case Stokes theorem
Z
(B) da =
I
B dl =
0
Z
J da
Now,
R
J da is the total current passing through the surface, which we call I
enc
, thus
I
B dl =
0
I
enc
This is the integral version of Amperes law; it is generalized to arbitrary steady currents. Just as the
Biot-Savart law plays a role in magnetostatics that Coulombs law assumed in electrostatics, so Amperes
plays the role of Gausss:
Electrostatics: Coulomb Gauss
Magnetostatics: Biot-Savart Ampere
In particular, for currents with appropriate symmetry, Amperes law in integral form offeres a lovely
and extraordinarily efcient means for calculating the magnetic eld.
85
5.4.3 Comparison of Magnetostatics and Electrostatics
The divergence and curl of the electrostatic eld are
E =

0
, Gausss law
E = 0 no name
These are Maxwells equations for electrostatics. The divergence and curl of the magnetostatic eld
are
B = 0 no name
B =
0
I
enc
Amperes law
These are Maxwells equations for magnetostatics. Maxwells equations and the force law
F = Q(E+vB)
constitute the most elegant formulation of electrostatics and magnetostatics.
5.5 Magnetic Vector Potential
5.5.1 The Vector Potential
Just as E = 0 permitted us to introduce a scalar potential (V) in electrostatics,
E =V
so B = 0 invites the introduction of a vector potential A in magnetostatics
B = A
The potential formulation automatically takes care of B=0 (since the divergence of a curl is always
zero); there remains Amperes law
B =(A) =( A)
2
A =
0
J
if we let
A = 0
with this condition to A, Amperes law becomes
2
A =
0
J
This is nothing but Poissons equation. This is actually three Poissons equations, on for each cartesian
component. Assuming J goes to zero at innity, we can read off the solution
A(r) =

0
4
Z
J(r
)
r
86
5.5.2 Summary; Magnetostatic Boundary Conditions
Just as the electric eld suffers a discontinuity at a surface charge, so the magnetic eld is discontinous at
a surface current. Only this time it is the tangential that changes. We can write this as
B
above
B
below
=
0
(K n)
where n is the unit vector perpendicular to the surface, pointing outward. Like the scalar potential
in electrostatics, the vector potential is continuous across any boundary
A
above
= A
below
for A = 0 guarantees that the n ormal component is continuous, and A = B, in the form
I
A dl =
Z
B da =
means that the tangential components are continuous. But the derivative of A inherits the discontinuity
of B.
A
above
n

A
below
n
=
0
K
5.6 Multipole Expansion of the Vector Potential
If you want an approximation formula for the vector potential of a localized current distribution, valid at
distant points, a multipole expansion is in order. The vector potential of a current loop can be written as
A(r) =

0
I
4
n=0
1
r
n+1
I
(r
)
n
P
n
(cos
)dl
or more explicetly
A(r) =

0
I
4
1
r
I
dl
+
1
r
2
I
r
cos
dl
+
1
r
3
I
(r
)
2
3
2
cos
2
1
2
dl
+...
as in the multipole expansion of V, we call the rst term (which goes like 1/r ) the monopole term,
the second (which goes like 1/r
2
) dipole, the third quadrupole, and so on. It happens that the magnetic
monopole term is always zeror, for the integral is just the total vector displacement around a closed loop
I
dl
= 0
this reects that there are no magnetic monopoles in nature. In the absence of any monopole contribu-
tion, the dominent term is the dipole (except in the rare case where it too, vanishes)
A
dip
=

0
I
4r
2
I
r
cos
dl
=

0
I
4r
2
I
( r r
)dl
this integral can be written in a more illuminating way if we invoke Eq 1.108 from Grifths, with
c = r :
I
( r r
)dl
= r
Z
da
87
then
A
dip
(r) =

0
4
m r
r
2
(5.12)
where m is the magnetic dipole moment:
mI
Z
da = Ia
where a is the area vector. The magnetic eld of a pure dipole is easiest to calculate if we put m at the
origin and let it point in the zdirection
According to equation 5.10, the potential at point (r, , ) is
A
dip
(r) =

0
4
msin
r
2

and hence
B
dip
(r) = A =

0
m
4r
3
(2cos r +sin
)
88
Problem # 1 Polygonal electromagnets
a) A square loop of wire, with size 2a 2a lies in the xy-plane centered at the origin. Calculate the
eld everywhere on the z -axis produced by a current I through this loop.
We know that a line segment produces a magnetic eld a distance s away of
B(s) = N

0
I
4s
2sin = N

0
I
2s
sin
where N represents the number of segments, in our case this is just 4, so we nd
B(s) =
2
0
I
s
sin
we can see from the diagram that
sin =
a
r
=
a
a
2
+s
2
=
a
2a
2
+z
2
s =
a
2
+z
2
thus we nd
B(r) =
2
0
I
s
a
2a
2
+z
2
this gives us our magnetic eld as a function of r, We know that the horizontal components will cancel
by symmetry and we will be left with a vertical component in the z direction, i.e
B
z
= Bcos = B
a
s
=
2
0
I
a
2
(a
2
+z
2
)
2a
2
+z
2
given that
cos =
a
s
89
b) What is the line integral of the eld along the z -axis from z = to z = ? Could you have
expected this result based on Amperes law? Explain.
We must now nd out what the integral yields
Z

B
z
dz =
2
0
Ia
2
dz
(a
2
+z
2
)(2a
2
+z
2
)
1/2
we can make a substitution of the form
z =
2atan dz =
2asec
2
d
plugging this into the integral gives
2
0
Ia
2
Z
/2
/2
2asec
2
d
a
2
(1+2tan
2
)(2a
2
(1+tan
2
)
1/2
=
2
0
Ia
2
Z
/2
/2
secd
a
2
(1+2tan
2
)
we can now multiply both top and bottom by cos
2
to give
2
0
Ia
2
Z
/2
/2
secd
a
2
(1+2tan
2
)
cos
2
cos
2
=
2
0
Ia
2
Z
/2
/2
cosd
a
2
(2sin
2
+cos
2
)
knowing
cos
2
= 1sin
2
allows us to simplify to
2
0
Ia
2
Z
/2
/2
cosd
a
2
(1+sin
2
)
doing another substitution
k = sin dk = cosd
which gives us
2
0
Ia
2
Z
1
1
dk
a
2
(1+k
2
)
=
2
0
I
tan
1
(k)
1
1
=
0
I
thus
2
0
Ia
2
dz
(a
2
+z
2
)(2a
2
+z
2
)
1/2
=
0
I
This is the line integral from R to R where R , what about the line integral of the semi-circle?
We can solve this integral by understanding how the magnetic eld behaves far way from the z axis, we
know from the Biot-Savart law that the magnetic eld goes like
B
1
R
2
If we let R we know that the integral along the semi-circle will go to zero, this exactly what we
would expect from Amperes law, which just says that if you enclose a current with an innite loop you
should get the total current enclosed.
c)Now nd the magnetic eld at the center of a n-sided regular polygon, letting the perpendicular
distance from the center to any side be a. Show that the result approaches that for a circular loop of radius
a in the limit n .
90
We know that the solution for the magnetic eld of a line segment on a plane is given by
B(s) =

0
I sin
2s
now we need to nd a relationship between the angle and the number of sides of a polygon.
4 sides 4(2) = 2 =

4
5 sides 5(2) = 2 =

5
6 sides 6(2) = 2 =

6
n sides n(2) = 2 =

n
using this defenition for in the expression for the magnetic eld yields
B(s) =

0
I sin(/n)
2s
we also know that
lim
n
sin(/n) =

n
thus we nd
B(s) =

0
I
2ns
this is only for 1 side, for n sides we nd
B(s) = n
0
I
2ns
=

0
I
2s
which is the magnetic eld of a loop of current with radius s.
Problem # 2
91
A large parallel-plate capacitor with uniform surface charge on the upper plate and on the lower
is moving with a constant speed v, as shown in Figure 3.
a) Find the magnetic eld bewteen the plates and also above and below them.
We know from Amperes law
I
B da =
0
I
enc
where
I
enc
= vL da = 2L z
plugging this into Amperes law gives
B2L =
0
vL
thus the magnetic eld is given by
B =

0
v
2
thus
B =
0
v
2
y above top plate
B = +
0
v
2
y below top plate
and the bottom plate is analogous to the top plate except all the signs change due to the surface
charge
B = +
0
v
2
y above bottom plate
B =
0
v
2
y below bottom plate
combining our results we nd
B = 0 above and below the plates
B =
0
v y in between the plates
b) Find the magnetic force per unit area on the upper plate, including its direction.
We know that
F
mag
= q(vB) =
Z
(vB)da =
Z
(KB)da
where
K
t
= v x
and we also know that the force is caused by the bottom plate, thus
B =

0
v
2
y
so we nd that the cross product yields
KB =
x y z
v 0 0
0

0
v
2
0
=

0
2
v
2
2
z
92
thus the force per unit area is given by
F
mag
=
Z
(KB)da =

0
2
v
2
2
zA
F
mag
A
=

0
2
v
2
2
z
c) At what speed v would the magnetic force balance the electrical force?
We know that
F
e
= F
mag
but we also know
F
e
= q

2
0
F
mag
= q
0
v
2
2
thus
v
2
=
1
0
v =
1
0
= c
thus we have shown that these two forces would be equal at a velocity of the speed of light.
Problem # 3
Suppose a very long solenoid carries a current of 10 A and has 1000 turns per cm. Find the radial
force per unit length f on one turn of the winding. Careful: the magnetic eld for an innite solenoid
is discontinuous - how did we handle such discontinuities to determine the pressure on the surface of
a charged conductor? Such forces pose severe design constraints on the construction of large, typically
superconducting, electromagnets.
We know that the force per unit length is given as
f =
F
mag
L
and we can nd F
mag
using
F
mag
=
Z
(vB)dq =
Z
(vB)dl = (vB)da = (KB)da = I
Z
(dl B)
93
and since we know that K and B are orthogonal we nd that the force per unit length is
F
mag
L
= IB
we can nd the magnetic eld by using Amperes law, we also know that the magnetic eld runs
parallel to the axis and by the righthand rule we know that it points upward inside the solenoid and
downward outside. For the loop inside the solenoid (loop 2) we nd
I
B dl = BL =
0
nI
enc
=
0
nIL
for the loop outside (loop 1) we nd
I
B dl = BL =
0
nI
enc
= 0
due to the fact that the loop does not enclose any current. Thus we nd that the magnetic eld for this
solenoid is given by
B =
0
nI z inside B = 0 outside
we also know that the magnetic eld of an innite solenoid is discontinous at a surface current, which
is the same case as the discontinuity of the electric eld at a surface charge. This means that the magnetic
eld is just the average of the magnetic eld below and and the magnetic eld above,
B =
B
above
+B
below
2
=

0
nI
2
z
thus we nd that the force per unit length is
F
mag
L
=

0
nI
2
2
z
plugging in the constants yield
F
mag
L
= 6.28
N
m
Problem # 4
Cycloid motion
Grifths Ex. 5.2 obtains a cycloid solution for motion of a charged particle in crossed electric and
magnetic elds using the Lorentz force law in the laboratory frame. Read this example, and then solve the
same problem using an alternate approach:
94
a) From relativity, but in the limit of velocities small compared to the speed of light, if frame S moves
with velocity
c with respect to (laboratory) frame S
, the electric eld is transformed as

E
= E+
cB
Taking B along x and E along z, as in the text, and assuming E cB in magnitude, choose the simplest
so that E
vanishes.
We know that for E
to vanish,
cB =E, thus
cB =
x y z
x

y

z
cB 0 0
=E z
which gives
cB
y
z +cB
z
y =E z
so we nd
=
E
cB
y
b) The two frames coincide at t = 0. Using the Galilei transformation for velocities (valid for 1),
if the particle is at rest at t = 0 in frame S , what is its initial velocity in frame S
?
we know that the Galilei transformations for velocities can be expressed as
u
= uv
or
u
x
= u
x
x
c
u
y
= u
y
y
c
u
z
= u
z
z
c
but we know that
x
=
z
= 0
thus we nd that the initial velocity in the S
frame is
u
y
= u
y
y
c u
y
= 0
so
u
y
=
y
c
c) The magnetic eld transforms in this case as
cB
= cB
E
Because E cB and 1, approximate B to be equal to B. Now, in frame S
solve for the motion

of the particle. Using the Galilei transformation for positions, determine the motion of the particle in the
lab frame.
95
If we assume that B
= B than
E = 0 E = 0
so we can see that
F
mag
= q(v
)
and lets assume thatv
has velocity components

v
= x, y, z B
=B
x
, 0, 0
thus we nd
v
x y z
x y z
B 0 0
= B z yB y z
the magnetic force is now given as
F
mag
= q(B z yB y z) = ma = m( y y+ z z)
which reduces to
qB z = m y qB y = m z
we can dene the cylotron frequency as
=
qB
m
and we nd
y = z z = y
if we differentiate the rst to get
...
y
= z =
2
y
which becomes
...
y
2
y = 0
this has a general solution of the form
y
(t) = A
1
cost +A
2
sint +A
3
y = A
1
sint +A
2
cost
y = A
1
2
cost A
2
2
sint
...
y
= A
1
3
sint A
2
3
cost
and we can see that this solution works for our differential equation. Now to nd z
(t) we can use the

rst derivative of y
which yields\
z =

y = A
1
2
sint A
2
2
cost
z = A
1
cos t A
2
sint
z
(t) = A
1
sint +A
2
cost +A
4
now we have two expressions for the motion of the particle,
y
(t) = A
1
cost +A
2
sint +A
3
z
(t) = A
1
sint +A
2
cost +A
4
96
now we need to nd what these constants are using our boundary coditions
y
(0) = z
(0) = 0 boundary condtion #1

v
y
(0) = c y v
z
(0) = 0 boundary condtion #2
from these two boundary conditions we nd
0 = A
2
+A
4
A
2
=A
4
from 1
A
2
=
c
A
1
= A
3
= 0 A
4
=
c
thus we nd that the solutions that describe this motion are given by
y
(t) =
c
sint z
(t) =
c
cost +
c
Problem # 5
6) Along with the necessary addition made by Maxwell, Amperes law reads
B =
0
J+
1
c
2
E
t
Taking the divergence of this equation, show that electric charge is conserved globally. We thus see
that the conservation of charge is tied fundamentally with the validity of Maxwells equations.
taking the divergence of both sides of this equation yields
(B) =
0
J+
1
c
2
t
( E)
but we know that
(B) = 0 E =

0
this gives us
0
J+
1
c
2
t
= 0
but
c
2
=
1
0
and so we nd
J =
t
because the charge is conserved, whatever ows out through a surface must come at the expense of
whatever is inside. Therefore this is the mathematical statement of local charge conservation. This is
generally refered to the continuity equation.
Problem # 6 On the gyromagnetic ratio
A thin uniform donut, carrying a charge Q and mass M, rotates about its axis as shown in Figure 1.
97
a). Find the ratio of its magnetic dipole moment to its angular momentum. This is called the gyro-
magnetic ratio ( or magnetomechanical ratio).
We know that the magnetic moment is dened as
m = L
where is the gyromagnetic ratio and L is the angular momentum of the system. We also know that
the magnetic moment can be dened as
m = Ia = Ir
2
where I is the current and a is the area. The angular momentum of a hoop is given by
L = I
where I is now the moment of inertia of a hoop, which is given by I = mr
2
, thus we nd the angular
momentum to be given as
L = mr
2
we also know that the current can be expressed as

I =
Q
T
T =
2
I =
Q
2
putting all this together we nd that the gyromagnetic ratio is given by
=
m
L
=
Qr
2
2mr
2
=
Q
2m
b). What is the gyromagnetic ratio for a uniform spinning sphere? [This requires no new calculation;
simply decompose the sphere into innitesimal rings, and apply the result form part a).]
From part a) we can see that the gyromagnetic ratio does not depend on the radius and thus decom-
posing the sphere into innitesimal rings will still yield the same ratio. The gyrommagnetic ratio is only
dependent on the total charge Q of the system and the mass M. This can be expressed mathematically as
m
L
=

q
i
2
r
2
i

m
i
r
2
i

=
Q
2m
c). According to quantum mechanics, the angular momentum of a spinning electron is
1
2
, where
is Plancks constant. What, then is the electrons magnetic dipole moment, in Am
2
? [This semiclassical
98
value is actually of by a factor of almost exactly 2. Diracs relativistic electron theory got the 2 right,
and Feynman, Schwinger, and Tomonaga later calculated tiny further corrections. The determination of
the electrons magnetic dipole moment remains the nest achievment of quantum electrodynamics, and
exhibits perhaps the most stunningly precise agreement between theory and experiment in all of physics.
Incidentally, the quantity (e/2m), where e is the charge of the electron and m is its mass, is called the
Bohr magneton.]
since we know that the magnetic moment is dened as
m = L
and the angular momentum of a spinning electron is given by
L =
1
2
we nd the magnetic moment of an electron to be given as

m =
Q
4m
= 4.6110
24
A m
2
99
Chapter 6
Magnetic Fields in Matter
6.1 Magnetization
All magnetic phenomenom are due to electric charges in motion, if you could examine a piece of magnetic
material on an atomic scale you would nd tiny currents: electrons orbiting around the nucleus and elec-
trons spinning about their axes. For macroscopic purposes, these current loops are so small that we may
treat them as magnetic dipoles. Ordinarily, they cancel each other out because of the random orientation
of the atoms. But when a magnetic eld is applied, a net alignment of these magnetic dipoles occur, and
the medium becomes magnetically polarized, or magnetized.
Unlike electric polarization, which is almost always in the direction of the eld, some materials acquire
a magnetization parallel to B (paramagnets) and some opposite to B (diamagnets). A few substances
called f erromagnets, retain their magnetization even after the external eld has been removed.
6.2 Torques and Forces on Magnetic Dipoles
A magnetic dipole experiences a torque in a magnetic eld just as a electric dipole does in an electric eld.
Lets calculate the torque on a rectangular current loop in a uniform magnetic eld B.
We have centered the loop at the origin, and a tilt of angle from the z axis toward the y axis and the
magnetic eld B points in the z dircetion. The forces on the two sloping sides cancel(they tend to stretch
the loop but not rotate it). The forces on the horizontal sides are likewise equal and opposite(so the net
force on the loop is zero), but they do generate a torque:
N = aF sin x
The magnitude of the force on each of these segments is
F = IbB
100
and therefore
N = IabBsin x = mBsin x
or
N = mB (6.1)
where m = Iab = IA is the magnetic dipole moment of the loop. This gives the exact torque on any
localized current distribution, in the presence of a uni f ormeld; in a noninu f orm eld it is the exact
torque (about the center) for a per f ect dipole of innitesimal size. Notice that this is analogous to the
torque given in electrostatics. In particular, the torque is again in such a direction as to line the dipole
parallel to the eld. It is this torque that accounts for paramagnetism. Paramagnetism normally occurs
in atoms or molecules with an odd number of electrons, where the extra unpaired electron is subject to
the magnetic torque.
In a uni f orm eld, the net force on a current loop is zero:
F = I
I
(dl B) = I
I
dl
B = 0
the constant B comes outside the integral, and the net displacement
H
dl around a closed loop vanishes.
For an in f initesimal loop, with dipole moment m, in a eld B, the net force is
F = (mB) (6.2)
which is analogous to the its electric counterpart F =(pE).
6.3 Effect of a Magnetic Field on Atomic Orbits
Electrons not onl;y spin; they also revolve around the nucleus- for simpliciuty, lets assume the orbit is a
circle of radius R. Although technically this orbital motion does not constitute a steady current, in practice
the period T = 2R/v is so short that unless you blink awfully fast, its going to look like a steady current
I =
e
T
=
ev
2R
Accordingly, the orbital dipole moment (IR
2
) is
m =
1
2
evR z
(The minus sign accounts for the negative charge of the electron). Like any other magnetic dipole, this
one is subject to a torque (mB) when the atom is placed in a magnetic eld. The orbital contribution to
paramagnetism is small. There is, however, a more signicant effect on the orbital motion: The electron
speeds up or slows down, depending on the orientation of B. For wheraa the centripetal acceleration v
2
/R
is ordinarily sustained by electric forces alone
1
4
0
e
2
R
2
= m
e
v
2
R
in the presence of a magnetic eld there is an additional force, e(vB). For the sake of argument, lets
say that B is perpendicular to the plane of orbit, as shown in gure 6.3, then
1
4
0
e
2
R
2
+e vB = m
e
v
2
R
101
Under these conditions, the new speed v is greater than v
e vB =
m
e
R
( v
2
v) =
m
e
R
( v +v)( vv)
or assuming the change v = v v is small
v =
eRB
2m
e
When B is turned on, the electron speeds up.
A change in orbital speed means a change in the dipole moment
m =
1
2
e(v)R z =
e
2
R
2
4m
e
B (6.3)
Notice that the change in m is opposite the direction of B. (An electron circling the other way would
have a dipole moment pointing upward, but such an orbit would be slowed down by the eld.
6.4 Magnetization and the Field of a Magnetized Object
In the presence of a magnetic eld, matter becomes magnetized. Upon a microscopic examination it will
be found to contain many tiny dipoles, with a net alignment along some direction. We describe the state
of magnetic polarization by the vector quantity
Mmagnetic dipole moment per unit volume
where M is called the magnetization; it plays a role analogous to the polarization P in electrostatics. In the
following section, we will not worry about how the magnetization got there- it could be paramagnetism,
diamagnetism, or even ferromagnetism- we shall take Mas given, and calculate the eld this magnetization
itself produces.
6.4.1 Bound Currents
Suppose we have a piece of magnetized material; the magnetic dipole moment per unit volume, M, is
given. What eld does this produce? Well, the vector potential of a single dipole m is given by Equation
5.12
A(r) =

0
4
m r
r
2
where r
is the seperation vector. In the magnetized object, each volume element d
carries a dipole
moment Md
, so that the total vector potential

A(r) =

0
4
Z
M(r
) r
r
2
d
That does it, in principle. But as in the electrical case, the integral can be cast in a more illuminating
form by exploiting the identity
1
r
r
2
102
with this,
A(r) =

0
4
Z
M(r
1
r
Integrating by parts, uisng product rule 7, gives

A(r) =

0
4
Z
1
r
M(r
)]d
M(r
)
r
We can write the latter as a surface integral

A(r) =

0
4
Z
1
r
M(r
)]d
0
4
Z
1
r
[M(r
) da
]
The rst term looks just like the potential of a volume current
J
b
=M (6.4)
while the second term looks like the potential of a surface current,
K
b
= M n (6.5)
where n is the normal vector. With these denitions,
A(r) =

0
4
Z
V
J
b
(r
)
r
+

0
4
Z
S
K
b
(r
)
r
da
What this means is that the potential (and hence also the eld) of a magnetized object is the same as
would be produced by a volume current J
b
= M throughout the material, plus a surface current K
b
=
M n,on the boundary.
6.5 The Auxiliary Field H
6.5.1 Amperes law in Magnetized Materials
In the previous section we found that the effect of magnetization is to establish bound currents J
b
=M
within the material and K
b
= M n on the surface. We are now ready to put everything together: the
eld attributable to bound currents, plus the eld due to everything else- which I shall conductocall free
current. The free current might ow through wires imbedded in the magnetized substance or, if the latter
is a conductor, through the material itself. In any event, the total current can be written as
J = J
b
+J
f
(6.6)
In view of Equation 6.4 and 6.6, Amperes law can be written
1
0
(B) = J = J
b
+J
f
= J
f
+(M)
or, collecting together the two curls
0
BM
= J
f
103
The quantity in paranthesis is designated by the letter H:
H
1
0
BM (6.7)
In terms of H, then, Amperes law reads
H = J
f
(6.8)
or, in integral form,
I
H dl = I
f
enc
(6.9)
where I
f
enc
is the total f ree current passing through the Amperian loop.
H plays a role analogous to D in electrostatics: Just as D allowed us to write Gauss
s law in terms of
the free charge alone, H permits us to express Ampere
s law in terms of the free current alone- and free

current is what we control. Bound current, like bound charge, comes along for the ride- the material gets
magnetized, and this results in bound currents
6.5.2 Boundary Conditions
The divergence of H is given by
H = M (6.10)
The magnetostatics boundary conditions can ne written in terms of H and the f ree current. From Equation
6.10 it follows that
H
above
H
below
=(M
above
M
below
)
while Equation 6.8 says
H
above
H
below
= K
f
n
In the presence of materials these conditions are sometimes more useful then the corresponding boundary
conditions on B
B
above
B
below
= 0
and
B
above
B
below
=
0
(K n)
6.6 Linear and Nonlinear Media
In paramagnetic and diamagnetic materials, the magnetization is sustained by the eld; when Bis removed,
M disappears. In fact, for most substances the magnetization is proprtional to the eld, provided the eld
is not too strong.
M =
m
H (6.11)
The constant of proportionality
m
is called the magnetic susceptibility; it is a dimensionless quantity that
variues from one substance to another- positive for paramagnets and negative for diamagnets. Materials
that obey the above relationship are called linear media. We also know
B =
0
(H+M) =
0
(1+
m
)H
104
for linear media. Thus B is also proportional to H
B = H
where

0
(1+
m
)
is called the permiability of the material. In vacuum, where there is no matter to magnetize, the suscep-
tibility
m
vanishes, and the permiability is
0
. Thats why
0
is called the permiability of free space.
Problem # 1 Forces on magnetic dipoles
A uniform current density J = J
0
z lls a slab straddling the yz plane, from x = a to x = +a. A
magnetic dipole m = m
0
x is situated at the origin.
a). Find the force on the dipole , using Equation 1.
The force on a magnetic diple is given as
F = (m B) (6.12)
We need to use Gausss law and an Amperean loop to nd what the magnetic eld is, we can see that
the magnetic eld will point in the y dircetion, We know it cannot point in the z direction or the x direction.
Z
B dl =
0
Z
J da
we know that dl = y y and da = xy z, thus
B =
0
J
0
x y
105
and so we nd that the force is
F = (m B) = 0
because m and B are orthogonal.
b). Do the same for the dipole pointing in the ydirection: m = m
0
y.
We nd that the force with this new dipole orientation is given by
F = (m B) =(m
0
y
0
J
0
x y) =

x
(m
0
0
J
0
x) x = m
0
0
J
0
x
c). In the elesctrostatic case the expression F = (p E) and F = (p )E are equivalent (prove it),
but this is not the case for the magnetic analogs (explain why). As an example calculate (m )B for the
congurations in (a) and (b).
We know that in electrostatics this statement is true
(p E) = (p )E
we knowthat we can expand the left hand term using Equation 4 in the from of Grifths, suing product
rules, i.e
(p E) = p(E) +E(p) +(p )E+(E )p
but we know
p(E) = E(p) = (E )p = 0
the rst term goes to 0 by denition, if E is a uniform constant electric eld, the second term goes to
0 because p is also a contstant vector, in a linear dialectric medium p is proportional to E. The last term
also goes to 0 because p is a costant vector, and when you take the derivative of a constant vector you get
0. Thus
(p E) = (p )E QED
The magnetic analog is given as
(m B) = m(B) +B(m) +(m )B+(B )m
B(m) = (B )m = 0
and
m(B) = 0
thus we nd
(m B) = m(B) +(m )B
for the congurations in a) we nd
(m )B = m
0
x
(
0
J
0
x) y = m
0
0
J
0
y
and for the conguration in b) we nd
(m )B = m
0
y
(
0
J
0
x) y = 0
106
The next set of problems concern the magnetic trapping of ultracold atoms, specically atoms of
87
Rb
(one of the rubidium isotopes). In one of their internal states (the F = 1 hyperne state), such atoms have
a magnetic dipole moment of m = m
B
/2 where m
B
= e/2m
e
is the Bohr magneton. Magnetic trapping
is achieved under the condition that the atomic magnetic moment remains anti-aligned with the magnetic
eld as the atom moves around in space, thus producing a trapping potential of the form U(r) = m|B|.
Problem # 2 Spherical quadrupole magnetic trap
First, lets consider a simple kind of magnetic trap known as the spherical quadrupole magnetic trap.
a) Consider two electromagnet coils sharing the same axis and separted by d, each of radius a and
N turns, each coil carrying current I running in the opposite direction as that of the other coil. Placing
the origin at the midpoint between the two coils, calculate the magnetic eld B(x, y, z) in the region near
the origin, keeping just the lowest order terms. Note that to determine the eld for radial displacements
(i.e. in the x or y directions), you can consider the symmetry of the problem and apply the condition of
zero-divergence for the magnetic eld.
An expansion about the origin of the magnetic eld B(x, y, z) yields
B
z
(x, y, z) = B
0
+ f
(B
z
) = ax +by +cz
B
y
(x, y, z) = B
0
+ f
(B
y
) = a
x +b
y +c
z
B
x
(x, y, z) = B
0
+ f
(B
x
) = a
x +b
y +c
z
we know that B
0
is 0 at the origin. We also know that
a = b = b = a
= c
= b
= c
= 0
thus we can write the expanded eld as
B(x, y, z) = a
x +b
y +cz
and since we know that B = 0 we nd
x
a
(x) +

y
b
(y) +

z
c(z) = 0
but since we know that this conguration has rotational symmetry we can say
x
(x) =

y
(y) a
= b
107
we nd
c =2a
now we just need to nd out what c is. From our expansion we can see that
f
(B
z
) = cz
and we know that the magnetic eld from a loop of wire at some point z away from the loop is given
by
B(z) =

0
I
2
a
2
(a
2
+z
2
)
3/2
but for our conguration, and the fact that we have two loops with currents in opposite directions we
nd
B(z) =

0
I
2
a
2
(a
2
+(z d/2)
2
)
3/2
a
2
(a
2
+(z +d/2)
2
)
3/2
and we know that

f
(B(z)) =

0
I
2
3a
2
(z d/2)
(a
2
+(z d/2)
2
)
5/2
+
3a
2
(z +d/2)
(a
2
+(z +d/2)
2
)
5/2
but since we want to know what the magnetic eld is at the origin we nd
f
(B(z = 0)) =
3
0
Ia
2
2
d
(a
2
+(d/2)
2
)
5/2
thus we nd that
c =
3
0
Ia
2
2
d
(a
2
+(d/2)
2
)
5/2
and
a
= b
=
c
2
=
3
0
Ia
2
4
d
(a
2
+(d/2)
2
)
5/2
thus the magnetic eld at the origin is given by

B(x, y, z) =
3
0
INa
2
4
d
(a
2
+(d/2)
2
)
5/2
(x x+y y2z z)
b) You should nd a spherical quadrupole eld with eld gradients in the radial and axial directions
determined by properties of the coils which create the eld. A spherical quadrupole eld can be written as
B(x, y, z) = B
(x x+y y2z z)
where B
is the radial eld gradient. Now put in some real numbers. Consider coils of radius a = 2.5
cm separated by d = 2a/
3 (the optimal separation), each of 20 turns running I = 92 amperes. What are

the axial and radial gradients (G/cm)?
We know that the radial eld gradient is given by
B
radial
=
3
0
INa
2
4
d
(a
2
+(d/2)
2
)
5/2
= 1561G/cm
108
and the axial eld gradient is
B
axial
=
3
0
INa
2
2
d
(a
2
+(d/2)
2
)
5/2
= 3120G/cm
Problem # 3 Majorana losses
In the spherical quadrupole magnetic eld described above, the potential U(r) = m|B| is clearly mini-
mized at a point, and, thus, particles with their magnetic moment anti-aligned with the eld will be trapped
by such a eld. However, a problem with the spherical quadrupole trap is the fact that it is always lossy.
Magnetic trapping requires that the atomic magnetic moment always follow the local orientation of the
magnetic eld. If the eld is changing its direction faster than an atoms orientation can keep up, the
atomic orientation may change, causing the atom to be ejected from a magnetic trap. This type of loss is
given the name Majorana loss.
a) Consider a
87
Rb atom moving with velocity vand impact parameter b in the xy plane of a spherical
quadrupole trap. Give an estimate for the critical impact parameter b
0
below which the atom has a high
probability of being lost, and above which the atom has a low probability of being lost. To do this, consider
the maximal rate at which the orientation of the eld is changing - this quantity has units of s
1
. Compare
this with the Larmor precession frequency of the atom in the local magnetic eld (this frequency is zero at
the origin and larger elsewhere).
We know that the orbital frequency and Larmar frequency for this problem are given by
orb
=
v
b
0
Lar
= B
setting these two expressions equal to each other yields
v
b
0
= B =B
b
0
but we know that
=
Q
2m
thus we nd
b
0
=
2mv
QB
b) Now we will estimate the Majorana loss rate for atoms at a certain temperature trapped in a spherical
quadrupole trap. To do this, consider that the treatment above determines an ellipsoidal volume within
which atoms at a given velocity may be lost from the trap. A loss rate can be calculated as the product nv
where n is the density of the atomic gas, is the area of the ellipsoid (which differs slightly if approached
from different directions; just take some characteristic area), and v is the typical velocity of atoms. To
estimate the volume occupied by the gas (youll need this to determine the density n), consider how large
is the volume in which the trap potential U(r) is smaller than k
B
Twhere k
B
is the Boltzmann constant.
Similarly, to obtain a typical velocity, consider the velocity at which the kinetic energy is k
B
T . Having
made all these estimates, nd the loss rate per atom given the temperature T and the radial eld gradient
B
.
Since we know that the lost rate is given by
L
rate
= nv
109
and we know that the energy is given by
U(r) = m|B| = m|B
(x x+y y)| = mB
r
and if the thermal energy is equal to the potential energy we get
k
B
T = mB
r
rearranging this for r gives
r =
k
B
T
mB
and the volume is given by

V =
4
3
r
3
4
3
b
3
0
but since
b
0
r
we nd the volume to be given as
V =
4
3
r
3
and using our expression for r we nd that the volume is given by
V =
4
3
k
B
T
mB
3
and we know that the loss rate is given by
L
rate
= nv =
N
V
v
we know that the velocity can be found using
1
2
mv
2
= k
B
T v =
2k
b
T
m
and also
= b
2
0
thus
L
rate
=
3b
2
0
2k
b
T
m
4
k
B
T
mB
3
=
3N
B
3
(B
)
2
Q(k
B
T)
2
c) Again, lets put in some numbers. Consider a gas of
87
Rb trapped in a spherical quadrupole trap
with a radial gradient of 120 G/cm. What is the Majorana loss rate per atom at temperatures of 1, 10, and
100 K.
Using
L
rate
N
=
3
3
B
(B
)
2
Q(k
B
T)
2
110
we nd
L
rate
= 3743
atoms
s
T = 1K
L
rate
= 37.4
atoms
s
T = 10K
L
rate
= 0.374
atoms
s
T = 100K
Problem # 4 The TOP trap
Losses from spherical quadrupole traps made evaporative cooling of atoms in magnetic traps ineffec-
tive until some manner was employed to avoid these losses. As you found above, the problem becomes
ever worse as the temperature of the gas gets lower, thus precluding the ultra-low temperatures needed
to reach Bose-Einstein condensation. One solution, employed by the team of Cornell and Wieman in
Colorado for the rst achievement of Bose-Einstein condensation in a gas, was the TOP (time-averaged
orbiting potential) trap. Consider that a spherical quadrupole trap is formed, and then an additional bias
eld is applied of the form
B
TOP
(x, y, z) = B
0
cost x+B
0
sint y
Atoms held in a time-varying magnetic trap of this type will experience a time-averaged potential at
every point so long as (1) the eld rotates at a rate faster than the motional time-scale of atoms so that an
atom experiences a local average of the magnitude of the eld, and (2) the eld rotates at a rate slower than
the Larmor precession frequency so that atomic magnetic moment remains anti-aligned with the magnetic
eld.
a) Assume that these conditions are met. What is the time-averaged magnitude of the magnetic eld
near the origin? You should nd that this magnitude has a minimum at the origin, and varies quadratically
away from the origin with a xed ratio of trap curvatures in the different directions.
We know that the total magnetic eld is given by
B
tot
= B
(x x+y y2z z) +B
0
cost x+B
0
sint y
= (B
x +B
0
cost) x+(B
y +B
0
cost) y2B
z z
and the magnitude is given by
B
tot
=
(B
x)
2
+(B
y)
2
+(2B
z)
2
+2B
B
0
(xcost +ysint)
thus magnitude of the time-averaged magnetic eld is given by
B
tot
=

2
Z
2/
0
(B
x)
2
+(B
y)
2
+(2B
z)
2
+2B
B
0
(xcost +ysint)dt
This can also be written as
B
tot
=

2
Z
2/
0
B
0
1+dt
where is given as
=
B
0
2
(x
2
+y
2
+4z
2
) +
2B
B
0
(xcost +ysint)
111
we can now Taylor expand about the origin to nd
1+ = 1+

2
2
8
+...
and we know that
2
=
B
0
2
(x
2
+y
2
+4z
2
) +
2B
B
0
(xcost +ysint)
2
and we know that the only thing that survives from this in the time averaged eld is
2B
B
0
(xcost +ysint)
2
where the rst higher order terms in x, y, and z vanish in
2
. We can write the time averaged expanded
eld as (after integration)
B
tot
= B
0
+
B
2
2B
0
(x
2
+y
2
+4z
2
)
B
2
4B
0
x
2
B
2
4B
0
y
2
b) Now put in some numbers. Suppose on top of the 120 G/cm radial gradient spherical quadrupole
eld, the time-varying TOP eld has a magnitude of 10 G and rotates at a rate of = 27.5 kHz. What
are the oscillation frequencies of a
87
Rb atom in this trap? (Note: if a one dimensional potential has the
form U(x) = (1/2)m
2
x
2
, then a particle of mass m will oscillate harmonically in this potential with
oscillation angular frequency ). Do the eld settings chosen satisfy the two conditions for adiabaticity
described above?
We know that the potential energy in three dimensions can be written as
U(r) =
1
2
m
2
r
2
=
B
B
where B is the time averaged magnetic eld from part a). Thus we nd that the oscillation frequency
can be expressed as
x
=
B
0
B
B
0
2m
Rb
= 151
1
s
y
=
B
0
B
B
0
2m
Rb
= 151
1
s
z
=
B
0
4
B
B
0
m
Rb
= 428
1
s
c) Finally, the critical temperature T
c
for Bose-Einstein condensation for atoms in a harmonic trap is
determined by the geometric mean of trapping frequencies and the number of atoms N by the relation
N = (3)
k
B
T
c

3
where k
B
is the Boltzmann constant and is the Riemann Zeta function ((3) =1.202). This equation
can be interpreted as stating that once the number of quantum states available to a gas at a temperature
112
T
c
roughly equals the number of particles, multiple occupancy of quantum states occurs and quantum
statistical effects such as Bose-Einstein condensation ensue. Consider the above TOP trap with 2 10
4
atoms. What is the Bose-Einstein condensation transition temperature?
The critical temperature can be expressed as
T
c
=
N
1/3

k
B
(3)
1/3
where is dened as the geometric mean
= (
x
z
)
1/3
= 213
1
s
thus
T
c
41.6 nK
Problem # 5 Refrigerator magnets and atom mirrors
Most of you have probably seen large, exible refrigerator magnets. These are at magnetic materials,
up to a couple of inches on a side, that stick to an iron refrigerator door by inducing magnetization in the
door with the magnetic elds emanating from the surface of the magnet.
a) Suppose you have such a magnet with a circular shape, and that magnet is uniformly magnetized,
with a magnetization out of the plane. What is the magnetic eld on the surface of the magnet at its center,
as a function of the radius of the magnet? On the basis of this observation, explain why this is not how
refrigerator magnets are magnetized.
Since we are assuming that this is a ring of charge then we already know what the magnetic eld is,
we have seen this problem before, the magnetic eld of a ring of charge is given by
B(z) =

0
I
2
R
2
(R
2
+z
2
)
3/2
z
but since we are only concerned about the magnetic eld at z = 0 we nd that the magnetic eld is
given by
B(R) =

0
I
2R
z
This would not make a very good magnetic due to the fact that as the radius increases the magnetic
eld decreases. If we say that the thickness of the magnetic is h and also a magnetization of M we can
write the current as
I = Mh = K
b
h
and nd that the magnetic eld due to this conguration is given by
B(r) =

0
Mh
2R
z =

0
K
b
h
2R
z
and this eld has the same charactersistics. Thus if we have a magnet that is twice as big, the magnetic
eld will be twice as small.
The magnetization in such a refrigerator magnet is actually a periodic linear array of oppositely ori-
ented magnetization. Let us consider the eld produced by such a magnetization. Specically, consider a
113
magnet that lls the space d < z < 0, being innite in extent in the x and y directions. Let the magneti-
zation M be oriented in the plane of the magnet, varying as
M = Msinkx x
where M gives the maximumvalue of the magnetization and k the spatial wavevector of its modulation.
b) Show that this magnetization can be represented by bound surface currents at z = 0 (top layer) and
z =d (bottom layer).
We know that
K
b
= M n
so for the top surface
K
b
(z = 0) = M z = Msinkx( y) n = z
and at the bottom
K
b
(z =d) = M z = Msinkx( y) n = z
c) Consider just the eld due to the top layer of bound currents. In the region z = 0, there are no
currents, and therefore B = 0 and B = 0. Thus, we can dene a scalar magnetic potential V
m
for
that obeys
2
V
m
= 0 and B = V
m
. Considering these relations, the boundary condition at z , and
the y-independence of this problem, show that V
m
can be written as
V
m
(r) =V
0
cos(kx +)e
k|z|
where the refer to solutions for V
m
either above or below the layer of currents. The values of V
0
and
are still to be determined. Hint: recall solutions to the Laplace equation that are separated by Cartesian
coordinates.
We can solve this problem by using seperation of variables in cartesian coordinates
V
m
= X(x)Z(z)
we need to solve Laplaces equation
2
V
m
=

2
V
m
x
2
+

2
V
m
y
2
= 0
which becomes
2
X
x
2
=k
2
X(x)

2
Z
z
2
= k
2
Z(z) k 0
114
which is just a second order differential equation with a decaying solution and an oscillating solution,
i.e
X(x) = Acoskx +Bsinkx Z(z) =Ce
kz
+De
kz
and our boundary condition z V
m
0 gives
C = 0 Acos(kx) +Bsin(kx) = A
cos(kx +)
where we nd
X(x) = A
cos(kx +) Z(z) = De
kz
and we can nally write the scalar potential as
V
m
= V
0
cos(kx +)e
k|z|
above the surface
V
m
= V
0
cos(kx +)e
k|z|
below the surface
where V
0
absorbs A
and D.
d) Now obtain an expression for the magnetic elds from the top layer of currents, still containing
unknown quantities V
0
and . Use Amperes law to determine those quantities and thus to fully determine
the elds.
We know that the magnetic eld can be found with
B(z > 0) =V
m
which can also be written as
B(x, z > 0) =
V
m
x
x+
V
m
z
z
= kV
0
(sin(kx +)e
kz
x+cos(kx +)e
kz
z)
and for below the surface we know that the magnetic eld can be found with
B(z < 0) =V
m
which can also be written as
B(x, z < 0) =
V
m
x
x+
V
m
z
z
= kV
0
(sin(kx +)e
kz
x+cos(kx +)e
kz
z)
to nd what V
0
and is we can use the integral form of Amperes law, which is given by
Z
B dl =
0
I
enc
but we know that dl = dx x on the top of the loop and I
enc
= K
b
dx thus we nd, we know that the
verticle line segments cancel, thus we nd
2
Z
l
0
V
0
(ksin(kx +) xkcos(kx +) z) dx x =
0
Z
l
0
Msinkxdx
which becomes
2kV
0
Z
(sinkx +)dx =
0
M
Z
sinkxdx
115
thus we can see that
V
0
=

0
M
2k
= 0
thus the magnetic eld on the top layer is
B(x, z > 0) =

0
M
2
e
kz
(sin(kx) x+cos(kx) z)
B(x, z < 0) =

0
M
2
e
kz
e) So far you have determined the eld from the top layer. Now add the elds from the bound currents
on both layers of the magnet. What is the magnitude of the eld on the surface of the magnet (z = 0)?
We know that the contribution to the magnetic eld from the bottom layer is going to be equal and
opposite as for the top layer (current is running in oppoiste direction) than the magnetic elds for the
bottom layer are given by
B(x, z > 0) =
0
M
2
e
kz
B(x, z < 0) =
0
M
2
e
kz
To nd the total magnetic we must add these two
B
T
= B
b
+B
T
=

0
M
2
(sin(kx) xcos(kx) z)
0
M
2
(sin(kx) xcos(kx) z) = 0
Aside frombinding to refrigerators, such periodic magnetizations can also be used as mirrors for beams
of atoms with magnetic moments. See, for instance, Roach et al., Physical Review Letters 75, 629 (1995).
2)
Problem # 6 The magnetic scalar potential
Maxwells equations for the magnetic eld (Amperes law, in particular) clearly demonstrate that the
magnetic eld cannot be dened via a scalar potential. Further, there are no magnetic charges, and, hence,
no magnetic charge density. Nevertheless, such quantities are useful even though they are solely ctional.
Here we consider the use of the magnetic charge density
m
and the magnetic scalar potential V
m
as devices
to determine the elds from magnetized materials.
a) We start with the eld of the ideal magnetic dipole. In class, we found out that the electric eld of an
ideal electric dipole and the magnetic eld of an ideal magnetic dipole are quite similar in form, the only
difference being the singular eld at the position of the dipole. Considering the relation between these two
elds, argue why one can write the eld from a magnetic dipole as
B(r) =V
m
(r) +
0
M(r) (6.13)
and determine the magnetic scalar potential V
m
in this case.
We know that since electric properties are analogous to magnetic properties, then the defenition of the
magnetic dipole seems reasonable. We know that we can compare the rst term of the magnetic dipole to
the rst term of the electric dipole. The second term is neccassery so that (like in electrostatics) we can
116
account for the discontinuity at r = 0 just like in the case of electrostatics. This is only in the absence of
free currents.
Since we know that the magnetic eld and electric eld have similar characteristics, we can see that
the electric eld dipole can be expressed as
E
dip
=
p
4
0
r
3
(2cos r +sin
) +
p
3
0
(r)
we can compare the magnetic dipole to this dipole to nd
0
M(r) analogous to
p
3
0
(r)
and so
0
m
4r
3
(2cos r +sin
) analogous to
p
4
0
r
3
(2cos r +sin
)
thus we can see from this relationship that
V
m
=

0
m
4r
3
(2cos r +sin
)
and since we know that the scalar potential for the electric eld dipole is given by
V
dip
(r, ) =
1
4
0
r p
r
2
=
pcos
4
0
r
2
and the analogous formula for the magnetic dipole scalar potential is given by
V
m
(r, ) =

0
4
r m
r
2
=

0
mcos
4r
2
b) Now consider the eld produced by a magnetization M. Using Eq. 1, and following our treatment
of the electric eld from polarized objects, show that the magnetic scalar potential in this case would be
dened as
V
m
(r) =

0
4
Z
S
M
(r
)
|r r
|
da+

0
4
Z
V
M
(r
)
|r r
|
d
3
r
Here V denotes the volume of the magnetized object, S its surface (with normal vector n), and
M
=
M and
M
= M n.
Using the expression for our scalar potential
V
m
=

0
4
r m
r
2
r = r r
and the total potential is given by

V
m
=

0
4
Z
V
r M(r
)
r
2
d
observing that
1
r
=
r
r
2
117
we have
V =

0
4
Z
V
M
1
r
and integrating by parts and using product rule number 5, gives

V =

0
4
Z
V
M
r
Z
V
1
r
(
M)d
or using the divergence theorem

V =

4
I
S
1
r
M da
+

0
4
Z
V
1
r
(
M)d
and we can dene
M
=
M
M
= M n
thus we nd
V
m
(r) =

0
4
I
S
M
(r
)
|r r
|
da+

0
4
Z
V
M
(r
)
|r r
|
d
3
r
The reason why this analogy works so far is that there are, as of yet, no free currents in the systems
were considering. Now we add those currents back in, and we nd, of course, that the magnetic eld can
no longer be given just as the gradient of a potential.
c) Give expressions for the B and H elds in the presence of both a magnetization M and a free current
density J, still using the magnetic scalar potentials described in parts (a) and (b).
In the presence of free currents we must write the magnetic eld as
B(r) = V
m
+
0
M(r) +
0
H
where we know that
B =
0
M+
0
H =
0
(J
b
+J
f
) =
0
J
and we also know
H = J
f
this can also be written as
J =
1
0
(B) = J
b
+J
f
Problem # 7
In solving this problem, keep in mind the boundary conditions for B and H. On a historical note, such
cavities were considered in the early days of electromagnetic theory as operational ways of measuring the
B and H elds inside magnetic materials.
118
Suppose the eld inside a large piece of magnetic material is B
0
, so that H
0
= (1/
0
)B
0
M.
a) Now a small spherical cavity is hollowed out of the material. Find the eld at the center of the
cavity, in terms of B
0
and M. Also nd H at the center of the cavity, in terms of H
0
and M.
We know that the magnetic eld for this conguration is given by
B = B
0
B
where B is the magnetic eld induced by the magnetization of the material, which was given in
example 6.1
B
=
2
3
0
M
thus the magnetic eld is given as
B = B
0
2
3
0
M
and the H eld is given by
H = H
0
H
where
H
=
B
0
M =
1
3
M
thus we nd
H = H
0
+
1
3
M
b) Do the same for a long needle-shaped cavity running parallel to M.
We know that we cann treat this problem like a solenoid problem, where we have found that the
magnetic eld is given by
B
=
0
nI =
0
K
b
=
0
M
and so the magnetic eld due to this conguration is given by
B = B
0
B
= B
0
0
M
and the auxillary eld is given by
H = H
0
H
where
H
=
B
0
M = 0
thus
H = H
0
c) Do the same for a thin wafer-shaped cavity perpendicular to M.
We know that
B = B
0
B
but B
can be found by assuming this is a loop of wire, which we have solve in problem 1
B
=

0
K
b
h
2R
z
119
but
K
b
= M n = Msin
where if we take the limit that this thin wafer gets thinner
h 0 B
0
thus we nd
B = B
0
and for the H eld we nd
H = H
0
+M
120
Chapter 7
Electrodynamics
7.1 Electromotive Force
7.1.1 Ohms Law
To make a current ow, you have to push on the charges. How fast they move, in response to a given
push, depends on the nature of the material. For most substances, the current density J is proprtional to
the f orce per unit charge, f
J =f (7.1)
The proportionality factor (not to be confused with the surface charge) is an emperical constant that
varies from one material to another; its called the conductivity of the medium. For most purposes metals
can be regarded as perfect conductors with =.
In principle, the force that drives the charges to produce the current could be anything- chemical,
gravitational, or trained ants with tiny harnesses. For our purpose, its usually an electromagnetic force
that does the job. In this case Equation 7.1 becomes
J = (E+vB) (7.2)
Ordinarily, the velocity of the charges is sufciently small that the second term can be ignored
J =E (7.3)
Equation 7.3 is called Ohms law, though the physics behind it is really contained in Equation 7.1, of
which 7.3 is just a special case. Ohms law can also be written as
V = IR (7.4)
The constant of proportionality R is called the resistance; its a function of the geometry of the arrange-
ment and the conductivity of the medium between two electrodes. Resistance in measured in ohms : an
ohm is avolt per ampere.
For steady currents and uni f orm conductivity
E =
1
J = 0 (7.5)
and therefore the charge density is zero; any unbalanced charge resides on the sur f ace. It follows, in par-
ticular, that Laplaces equation holds within a homogeneous ohmic material carrying a steady current, so
121
all the tools and tricks of Chapter 3 are available for computing the potential. As a results of the collisions
between the electrons moving through a medium, the work done by the electric force is converted into
heat in the resistor. Since the work done per unit charge is V and the charge owing per unti time is I, the
power delivered is
P =VI = I
2
R (7.6)
This is called the Joule heating law. With I in amperes and R in ohms, P comes out in watts (joules per
second).
7.1.2 Electromotive Force
There are two forces that drive current around in a circuit: the source f
s
, which is ordinarily conned to
one portion of the loop (a battery, say), and the electrostatic force, which serves to smooth out the ow
and communicate the inuence of the source to distant parts of the circuit:
f = f
s
+E (7.7)
The physical agency responsible for f
s
can be any one of may different things. Whatever the mechanism,
its net effect is determined by the line integral of f around the circuit
E =
I
f dl =
I
f
s
dl (7.8)
(Because
H
E dl = 0 for electrostatic elds, it doesnt matter whether you use f or f
s
.) E is called the
electromotive force, or emf, of the circuit. An electromotive force can also arise if a loop of wire is
moving through a magnetic eld, in this case the emf is dened as
E =
d
dt
(7.9)
This is known as the ux rule for motional ems, where is the ux of B through the loop and is dened
as

Z
B da (7.10)
Apart from its delightful simplicity, it has the virtue of applying to nonrectangular loops moving in
arbitrary dircetions through nonuniform magnetic elds; in fact, the loop need not even maintain a xed
shape. The ux rule is a nifty short-cut for calculating motional emfs. It does not contain any new physics.
Ocassionally you will run across problems that cannot be handled by the ux rule; for these one must go
back to the Lorentz force law itself.
7.2 Electromagnetic Induction
In 1831 Micheal Faraday reported on a series of experiments, including three that can be characterized as
follows
Experiment 1. He pulled a loop of wire to the right through a magnetic eld. A current owed in the
loop
Experiment 2. He moved the magnet to the left, holding the loop still. Again current owed in the loop
122
Experiment 3. With both the loop and the magnet at rest, he changed the strength of the eld (he used
an electromagnet, and varied the current in the coil). Once again, current owed in the loop
The rst experiment, of course, is an example of motional emf, conveniently expressed by the ux rule
Equation 7.9. I dont think it will surprise you to learn that exactly the same emf arises in experiment
number 2, all that really matters is the relative motion of the magnet and the loop. Indeed, in light of
special relativity it has to be so. Faraday had an ingenious inspiration
A changing magnetic eld induces an electric eld
It is this induced electric eld that accounts for the emf in Experiment 2. Indeed, if (as Faraday found
emperically) the emf is again equal to the rate of change of the ux,
E =
I
E dl =
d
dt
(7.11)
then E is related to the change in B by the equation
I
E dl =
Z
B
t
da (7.12)
This is Faradays law, in integral form,. We can convert it to differential form by applying Stokes
theorem
E =
B
t
(7.13)
Note that Faradays law reduces to the old rule
H
E dl = 0 in the static case (constant B) as, of course, it
should.
In Experiment 3 the magnetic eld changes for entirely different reasons, but according to Faradays
law an electrif eld will again be induced, giving rise to an emf d/dt. Indeed, one can subsume all
three cases into a kind of universal ux rule:
Whenever (and for whatever reason) the magnetic ux through a loop changes, an emf
E =
d
dt
Will appear in the loop.
7.3 Inductance
Suppose you have two loops of wire, at rest. If you run a steady current I
1
around loop 1, it produces a
magnetic eld B
1
. Some of the eld lines pass through loop 2; let
2
be the ux of B
1
through 2. You
might have a tough time actually calculating B
1
, but at a glance at the Biot-Savart law
B
1
=

0
4
I
1
I
dl
1
r
2
reveals one signicant fact about the eld; it is proportional to the current I
1
. Therefore, so too is the ux
through loop 2:
2
=
Z
B
1
da
2
(7.14)
123
Thus
2
= M
21
I
1
(7.15)
where M
21
is the constant of proportionality; it is known as the mutual inductance of the two loops.
There is a cute formula for the mutual inductance, which can be derived by expressing the ux in terms
of the vector potential and invoking Stokes theorem
2
=
Z
B
1
da
2
=
Z
(A
1
) da
2
=
I
A
1
dl
2
the vector potential can be written as
A
1
=

0
I
1
4
I
dl
1
r
and hence
2
=

0
I
1
4
I
I
dl
1
r
dl
2
Evidently
M
21
=

0
4
I I
dl
1
dl
2
r
(7.16)
This is the Neumann formula; it involves a double liune integral. It is not very useful for practical
calculations, but it does reveal two important things about mutual inductance:
1. M
21
is a purely geometrical quantity, having to do with sizes, shapes, and relative positions of the
two loops.
2. The integral in Equation 7.15 is unchanged if we switch the roles of loops 1 and 2; it follows that
M
21
= M
12
= M
Suppose now that you vary the current in loop 1. The ux through loop 2 will vary accordingly, and
Faradays law says this changing ux will induce an emf in loop 2
E
2
=
d
2
dt
=M
dI
1
dt
(7.17)
What a remarkable thing: Everytime you change the current in loop 1, and induced current ows in loop
2- even though there are no wires connecting them.
Come to think of it, a changing current not only induces an emf in any nearby loops, it also induces
an emf in the source loop itsel f . Once again, the eld (and therefore also the ux) is proportional to the
current
= LI (7.18)
The constant of proportionality L is called the self-inductance (or simply the inductance) of the loop. As
with M, it depends on the geometry (size and shape) of the loop. If the current changes, the emf induced
in the loop is
E =L
dI
dt
(7.19)
Inductance is measured in henries (H); a henry is a volt-second per ampere.
124
7.3.1 Energy in Magnetic Fields
It takes a certain amount of energy to start a current owing in a circuit. The work done on a unit charge,
against the back emf, in one trip around the circuit is E (the minus sign records the fact that this is the
work done by you against the emf, not the work done by the emf. The amount of charge per unit time
passing down the wire is I. So the total work done per unit time is
dW
dt
=EI = LI
dI
dt
If we start with zero current and build it up to a nal value I, the work done is
W =
1
2
LI
2
(7.20)
It does not depend on how long we take to crank up the current, only on the geometry of the loop (in the
form of L) and the nal current I. We can also write the work done in terms of the magnetic eld as
W =
1
2
0
Z
all space
B
2
d (7.21)
In view of this result, we say the energy is stored in the magnetic eld, in the amount (B
2
/2
0
) per unit
volume.
7.4 Maxwells Equations
We are now ready to write down all of Maxwells equations (with corrections), they are summarized as
(i) E =

0
(Gausss law)
(ii) B = 0 (no name)
(iii) E =
B
t
(Faradays law)
(iv) B =
0
J+
0
0
E
t
(Amperes law with Maxwells correction)
(7.22)
Together with the force law
F = q(E+vB) (7.23)
they summarize the entire theoretical content of calssical electrodynamics (save for some special properties
of matter). Even the continuity equation
J =
t
(7.24)
which is the mathematical expression of conservation of charge, can be derived from Maxwells equations
by applying the divergence to number (iv). I have written Maxwells equation in the traditional way. I
think it is logically preferable to write them as
(i) E =

0
(Gausss law)
(ii) B = 0 (no name)
(iii) E+
B
t
= 0 (Faradays law)
(iv) B
0
0
E
t
=
0
J (Amperes law with Maxwells correction)
(7.25)
with the elds (E and B) on the left hand side and the sources ( and J) on the right. This notation
emphasizes that all electromagnetic elds are ultimately attributable to charges and currents. Maxwells
equatins tell you how charges produce elds, the force law tells you how elds affect charges.
125
7.5 Maxwells Equations in Matter
Maxwells equations in the form of 7.25 are complete and correct as they stand. However, when you are
working with materials that are subject to electric and magnetic polarization there is a more convenient
way to write them. In terms of f ree charge and currents they are summarized as follows
(i) D =
f
(iii) E =
B
t
(7.26)
(ii) B = 0 (iv) H = J
f
+
D
t
Some people regard these as the true Maxwells equations, but they are in no way more general than
7.25; they simply reect a convenient division of charge and current into free and non-free parts. And
they have the disadvantage of hybrid notation, since they contain both E and D, both B and H. They must
be supplemented, therefore, by appropriate constitutive relations, giving D and Hb in terms of E and B.
These depend on the nature of the material; for linear media
P =
0
e
E and M =
m
H
so
D =E and H =
B
where =
0
(1+
e
) and =
0
(1+
m
). Incidentally, youll remember that D is called the electric dis-
placement; thats why the second term in the Ampere/Maxwell equation (iv) is called the displacement
current.
7.6 Boundary Conditions
In general, the elds E,B,D, and H will be discontinuous at a boundary between two different media, or at
a surface that carries charge density or current density K. The explicit form of these discontinuities can
be deduced from Maxwells equations 7.27, in their integral form
(i)
I
S
D da = Q
f enc
(iii)
I
P
E dl =
d
dt
Z
S
B da (7.27)
(ii)
I
S
B da = 0 (iv)
I
P
H dl = I
f enc
+
d
dt
Z
S
D da
where (i) and (ii) are dened over any closed surface S, and (iii) and (iv) are for any surface S bounded by
the closed loop P. For (i) we nd
D
1
D
2
=
f
(7.28)
for (ii) we nd
B
1
B
2
= 0 (7.29)
for (iii) we nd
E
1
E
2
= 0 (7.30)
and for (iv) we nd
H
1
H
2
= K
f
n (7.31)
126
Equation 7.29-7.32 are the general boundary conditions for electrodynamics. In the case of linear
media, they can be expressed in terms of E and B alone
(i)
1
E
1

2
E
2
=
f
(iii) E
1
E
2
= 0 (7.32)
(ii) B
1
B
2
= 0 = 0 (iv)
1
1
H
2
H
2
= K
f
n
In particular, if there is no free charge or free current at the interface, then
(i)
1
E
1

2
E
2
= 0 (iii) E
1
E
2
= 0 (7.33)
(ii) B
1
B
2
= 0 = 0 (iv)
1
1
H
2
H
2
= 0
Problem # 1
Notice the following parallel:
D = 0 E = 0
0
E = DP
B = 0 H = 0
0
H = B
0
M
thus, the transcription D B, E H, P
0
M,
0
0
turns an electrostatic problem into an analo-
gous magnetostatic one. Use this observation together with your knowledge of the electrostatic results, to
rederive
a) The magnetic eld inside a uniformly magnetized sphere. (Eq. 6.16)
The magnetic eld of a uniformly magnetized sphere is given by Equatio 6.16
B =
2
3
0
M
The electric eld of a uniformly polarized sphere is given by Equation 4.14
E =
1
3
0
P
if we let
E H P
0
M
0
0

0
H = B
0
M
we nd
H =
1
3
M
0
H =
1
3
0
M
127
which simplies to
B
0
M =
1
3
0
M
thus
B =
2
3
0
M
b) The magnetic eld inside a sphere of linear magnetic material in an otherwise uniform magnetic
eld. (Prob. 6.18)
From example 4.7 we know that the electric eld inside a linear dialectric sphere is given by Equation
4.49
E =
3
2+
r
E
0
If we let
E H
r
= 1+
e
m
= 1+
m
E
0
H
0
thus we have
H =
3
2+
m
H
0
if we let
P =
0
e
E =
0
m
H =
0
M
therefore
M =
m
H
we also know
0
H = B
0
M H =
B
0
(1+
m
)
and now to nd what H
0
is we must use
0
H
0
= B
0
0
M = B
0
M = 0 outside
thus
H
0
=
B
0
0
and so we nd that the magnetic eld inside a sphere of linear magnetic material in an otherwise
uniform magnetic eld is given by
B =
3(1+
m
)
3+
m
B
0
c) The average magnetic eld over a sphere, due to steady currents within the sphere. (Eq. 5.89)
We know that the average magnetic elds over a sphere due to steady currents is given by Equation
5.89
B
ave
=

0
4
2m
R
3
we can derive this using our knowledge of what the average electric eld due to a similar electrostatic
problem. The average electric eld is given by Equation 3.105 as
E
ave
=
1
4
0
p
R
3
128
we also know that
P =
p
V
where we know that the volume is given by
V =
4
3
R
3
thus we can rewrite the average electric eld as
E
ave
=
1
3
0
P
which is identical to part a) of this problem which yielded
B
ave
=
2
3
0
M
but we know that this can also be written as
M =
m
V
B
ave
=
2
3V

0
m
where this can be written as
B
ave
=
2
0
4
m
R
3
Problem # 2
A metal bar of mass m slides frictionless on two parallel conducting rails a distance l apart (Figure 1).
A resistor R is connected across the rails and a uniform magnetic eld B, pointing into the page, lls the
entire region.
a) If the bar moves to the right at speed v, what is the current in the resistor? In what direction does it
ow?
Since we know that
I =
V
R
129
from Ohms law, where V is now the induced emf, which is given by
V =E =
d
dt
and is the magnetic ux which is given by
=
Z
B da = Bxl
where the normal are vector is into the page. The emf is given by
E =
dx
dt
Bl =Blv
thus the current is given by
I =
Blv
R
counter-clockwise
b) What is the magnetic force on the bar? In what direction?
We know that the Lorentz force is given by
F
B
= q(v B) = qvB
but since we know that
qv = Il
we nd
F
B
= IlB =
vl
2
B
2
R
( x)
there is a magnetic force in the x direction.
c) If the bar starts out with speed v
0
at time t = 0, and is left to slide, what is its speed at a later time t?
Since we know that
F
B
= ma = m
dv
dt
we can solve for the velocity, i.e
vl
2
B
2
R
= m
dv
dt
which can be written in a more suggestive way
Z
t
0
l
2
B
2
Rm
dt =
Z
v
v
0
dv
v
which is a simple integral with a solution of
v = v
0
e
kt
k =
l
2
B
2
Rm
d) The initial kinetic energy of the bar was, of course,
1
2
mv
2
0
. Check that the energy delivered to the
resistor is exactly
1
2
mv
2
0
.
130
We know that the energy is given by
E =
Z
Pdt
where P is the power that is obtained from the loss of kinetic energy of the bar, the power is dened as
P = I
2
R =
l
2
v
2
B
2
R
=
l
2
B
2
R
v
2
0
e
2kt
where k has been previously dened, so we nd that the total energy delivered to this transistor is
E =
l
2
B
2
R
Z

0
e
2kt
dt =
l
2
B
2
R
Rm
2l
2
B
2
v
2
0
=
1
2
mv
2
0
Problem # 3
Two tiny wire loops, with areas a
1
and a
2
, are situated a displacement r apart (Figure 2).
a) Find their mutual inductance. [Hint : Treat them as magnetic dipoles, and use Eq. 5.87.] Is your
formula consistent with Eq. 7.23?
Using Equation 5.87
B
dip
=

0
4
1
r
3
[3(m r) r m]
using Biot-Savart law and the denition of the ux thru loop 2 from the magnetic eld due to loop 1
2
=
Z
B
1
da
2
we can dene the mutual inductance of the two loops as
2
= M
21
I
1
where M
21
is the mutual inductance. Using this we can now nd
M
21
=

2
I
1
=
1
I
1
Z
B
1
da
2
given the denition of the magnetic eld due to a dipole we nd
M
21
=
1
I
1
0
4
1
r
3
[3(m
1
r) r m
1
] a
2
131
knowing
m
1
= a
1
I
1
m
2
= a
2
I
2
gives us
M
21
=

0
4
1
r
3
[3(a
1
r)(a
2
r) a
1
a
2
]
we can also show that the is the same if we look at it from the perspective of the other loop,
M
12
=

1
I
2
=
1
I
2
Z
B
2
da
1
given the denition of the magnetic eld due to a dipole we nd
M
12
=
1
I
2
0
4
1
r
3
[3(m
2
r) r m
2
] a
1
knowing
m
1
= a
1
I
1
m
2
= a
2
I
2
gives us
M
21
=

0
4
1
r
3
[3(a
2
r)(a
1
r) a
2
a
1
]
which is excatly the same, thus we can conclude
M
21
= M
12
= M
b) Suposse a current I
1
is owing in loop 1, and we propose to turn on a current I
2
in loop 2. How
much work must be done, against the mutually induced emf, to keep the current I
1
owing in loop 1? In
light of this result, comment on Eq.6.35.
Since we know that the total work done per unit time is given by
dW
dt
=EI
where E is the induced emf and I is the current going thru the loop. We need to nd
dW =E
1
I
1
dt
but we know that the induced emf on 1 is given by
E
1
=
d
1
dt

1
=
Z
B
2
da
1
= MI
2
and so we nd
E
1
=M
dI
2
dt
putting this into our equation for the work yields
dW = MI
1
dI
2
W = MI
1
I
2
We found the mutual inductance from part a), substituting this in gives us
W =

0
4
1
r
3
[3(a
1
I
1
r)(a
2
I
2
r) a
1
I
1
a
2
I
2
]
132
where we have dened
m
1
= a
1
I
1
m
2
= a
2
I
2
which gives us
W =

0
4
1
r
3
[3(m
1
r)(m
2
r) m
1
m
2
]
this looks very similar to the interaction energy of two magnetic dipoles.
W =U
where U is the interaction energy of two magnetic dipoles
U =

0
4
1
r
3
[m
1
m
2
3(m
1
r)(m
2
r)]
Problem # 4
The current in a long solenoid is increasingly linear in time, so that the ux is proportional to t :=t.
Two voltmeters are connected to diametrically opposite points (A and B), together with resistors (R
1
and
R
2
), as shown in Figure 3. What is the reading at each voltmeter? Assume that these are ideal voltmeters
that draw negligable current (they have huge internal resistance), and that a voltmeter registers
R
b
a
E dl
between the terminals and through the meter. [Answer : V
1
= R
1
/(R
1
+R
2
); V
2
= R
2
/(R
1
+R
2
).
Notice that V
1
=V
2
, even though they are connected to the same points!]
Since we know that these resistors are in series we can write the current as
I =
V
ind
R
T
=
V
ind
R
1
+R
2
but we know that the induced emf is given by
V
ind
=E =
d
dt
=
thus the current running thru this loop is given by
I =

R
1
+R
2
We can nd what the potential is across resistor 1 using
V
1
=
Z
a
b
E dl =IR
1
=
R
1
R
1
+R
2
133
and to nd the potential across resistor 2 we use
V
2
=
Z
b
a
E dl = IR
2
=
R
2
R
1
+R
2
thus
V
1
=
R
1
R
1
+R
2
V
2
=
R
2
R
1
+R
2
Problem # 5
Suppose
E(r, t) =
1
4
0
q
r
2
(vt r) r; B(r, t) = 0
(the theta function is dened in Prob. 1.45b). Show that these elds satisfy all of Maxwells equations,
and determine and J. Describe the physical situation that give rise to these elds.
If we look at
E =

0
Using product rule # 5
( f A) = f ( A) +A (f )
where f =(vt r) and A =
r
r
2
, thus we nd
E =
q
4
0
(vt r)
r
r
2
+
r
r
2
(((vt r))
0
Equation 1.99 in Grifths gives
r
r
2
= 4
3
(r)
and
r
r
2
(((vt r)) =
r
r
2

r
((vt r)
=
r
r
2
d
d(vt r)
d(vt r)
dr
=
(vt r)
r
2
Since we know that
d
d(vt r)
= (vt r)
d(vt r)
dr
=1
Thus we nd that
=q
(vt r)
3
(r)
(vt r)
4r
2
The second Maxwell equation is

B = 0 B = 0
The third equation is
E =
B
t
Since B = 0, we nd
E = 0
134
For the last Maxwell equation
B =
0
J+
0
0
E
t
We know that
B = 0
Thus
J =
0
E
t
We can nd
E
t
=
q
4
0
r
2
t
(vt r) =
q
4r
2
d
d(vt r)
d(vt r)
dt
=
q
4r
2
v(vt r)
Thus
J =
q
4r
2
v(vt r)
Problem # 6
Assuming that Coulombs law for magnetic charges (q
m
) reads
F =

0
4
q
m
1
q
m
2
r
where r
is Grifths script r, which is the seperation vector. Work out the force law for a monopole q
m
moving with velocity v through electric and magnetic elds E and B.
Looking at the paper written in 1961 by Robert Katz fromKansas state university we nd that Maxwells
equations can be written as
D = E =
B
t
+J
m
B =
m
H =
D
t
+J
he has dened the following properties

D =
q
4r
2
r B =
q
m
4r
2
r H =
(qv r)
4r
2
E =
(q
m
v r)
4r
2
set up in this way magnetic poles satisfy the continuity equation; we demend the conservation of poles.
We can immideatly write the magnetic analog of the Lorentz force equation as
F
m
= q
m
[HvD] = q
m
(BvB) = q
m
(B
1
c
2
vE)
where the 1/c
2
was added in order to make the units work out, and also due to the motivation that E = cB.
We have shown why these are the units needed in the next part of this explanation.
We claim to nd this using simple unit analysis and Equations 7.43 from Grifths. We know that the
Lorentz force for electric charges is given by
F
e
= q
e
[E+vB]
135
Thus by analogy we should nd
F
m
= q
m
[Bc
1
(vE)]
we need to nd a constant c
1
that would make the units work out. What are the units of q
m
? We know that
the units of the force must be in Newtons and we know that the units of the magnetic eld are
B
N
Am
F
m
= q
m
B
where the units of q
m
must be (Am). Now looking at the second term we nd
F
m
= q
m
c
1
(vE)
If
q
m
Am vE
Nm
As
2
Thus the units of c
1
must be
c
1
s
2
m
2
In order to have the force be in Newtons, a good choice for this units would be the speed of light c, thus
F
m
= q
m
[B
1
c
2
vE]
now for the sign we must look at Equations 7.43 in Grifths, we can see that
E =
0
J
m
B
t
B =
0
J
e
+
0
0
E
t
we can see that when B E we gain a minus sign and thus we can conclude that
F
m
= q
m
[B
1
c
2
vE]
Problem # 7
In a perfect conductor, the conductivity is innite, so E=0 (for J = E), and any net charge resides
on the surface (just as it does for an imper f ect conductor, in electrostatics).
a) Show that the magnetic eld is constant (B/t = 0), inside a perfect conductor.
We know that
E =
B
t
Since we know that E = 0 we nd
B
t
= 0
This will only be true if the magnetic eld is constant.
b) Show that the magnetic ux through a perfectly conducting loop is contant.
We know that
E =
Z
E dl =
d
dt
= 0
136
Because E = 0, and this will only be true if is a constant.
A superconductor is a perfect conductor with the additional property that the (constant) B inside is
in fact zero. (This ux exclusion is known as the Meissner effect.)
c) Show that the current in a superconductor is conned to the surface.
Using Maxwells equation
B =
0
J+
0
0
E
t
= 0
and since we know that E = 0 than
J = 0
and any charges that are moving must reside on the surface.
d) Superconductivity is lost above a certain critical temperature (T
c
), which varies from one material
to another. Suppose you had a sphere (radius a) above its critical temperature, and you held it in a uniform
magnetic eld B
0
z while cooling it below T
c
. Find the induced surface current density K, as a function of
the polar angle .
Using Example 5.11 in Grifths we nd that the magnetic eld inside of a rotating magnetized sphere
is given by
B =
2
3
0
r z
This is the electric eld before the temperature falls below the critical temperature T > T
c
. Once we
decrease the temperature below the critical temperature we know that the magnetic eld inside the sphere
must be zero due to the Meissner effect and thus we know that we must induce a surface current that
will oppose the external eld. Thus
B
external
= B
0
z B
induced
=B
0
z
Thus we know that
B =
2
3
0
r z =B
0
z r =
3
2
B
0
0
and since we know that the surface current is dened as
K =(r) =rsin
and using the result above we nd

K =
3
2
B
0
0
sin
Problem # 8
A familiar demonstration of superconductivity is the levitation of a magnet over a piece of supercon-
ducting material. This phenomenom can be analyzed using the method of images. Treat the magnet as
a perfect dipole m, a height z above the origin (and constrained to point in the z direction), and pretend
that the superconductor occupies the entire half-space below the xy plane. Because of the Meissner effect,
B = 0 for z 0, and since B is divergenceless, the normal (z) component is continuous, so B
z
= 0 just
above the surface. This boundary condition is met by the image conguration in which an identical dipole
is placed at z, as a stand-in for the superconductor; the two arrangements therefore produce the same
magnetic eld in the region z > 0.
137
a) Which way should the magnetic dipole point (+z or z )?
We know that the image dipole must point in the z direction so that at z = 0 the perpendicular
component of the magnetic eld must be 0. Mathematically
m
1
= m
1
z m
2
=m
2
z
where m
1
is the original dipole and m
2
is the image dipole.
b) Find the force on the magnet due to the induced currents in the superconductor (which is to say, the
force due to the image dipole). Set it equal to M
g
(where M is the mass of the magnet) to determine the
height h at which the magnet will oat. [Hint : refer to Prob 6.3]
We know that the force on a dipole due to a magnetic eld is dened as
F = (m B)
since this is an image problem we know that this can also be written as
F =(m
1
B
2
)
which says that there is a force on the original dipole to to the magnetic eld from the image dipole, which
is dened as
B
2
=

0
4r
3
[3(m
2
r) r m
2
]
we know that
m
1
= m
1
z m
2
=m
1
z r = z
we can now write the magnetic eld from the image dipole as
B
2
=

0
4r
3
[3(m
1
z z) z+m
1
z] =

0
4r
3
[2m
1
z]
Thus we nd the force to be given as
F =
0
2
m
2
1
r
3
Thus we nd
F =
3
0
m
2
1
2r
4
we know that
r = 2z
which is just the speration distance between the two dipoles, thus the force is given by
F =
3
0
m
2
1
32z
4
= M
g
If we let z = h we nd that the height is found to be
h =
3
0
m
2
1
32M
g
1/4
138
c) The induced current on the surface of the superconductor (the xy plane) can be determined from the
boundary condition on the tangential component of B (B =
0
(K z). Using the eld you get from the
image conguration, Show that
K =
3mrh
2(r
2
+h
2
)
5/2

where r is the distance from the origin.

We can nd what the induced current if we know what the magnetic eld is on the xy plane, thus we
have to nd what the total magnetic eld is at z = 0 this can be achieved by guring out what the magnetic
eld is due to the two dipoles, we know that the magnetic eld from a dipole is given by
B
dip
=

0
4r
3
[3(m r) r m]
thus for dipole 1 which I will consider my real dipole we nd
B
1
=

0
4r
3
1
[3(m
1
r
1
) r
1
m
1
]
and like wise for the image dipole
B
2
=

0
4r
3
2
[3(m
2
r
2
) r
2
m
2
]
thus we know that the total magnetic eld is given by
B
T
= B
1
+B
2
=

0
4
1
r
3
1
[3(m
1
r
1
) r
1
m
1
] +
1
r
3
2
[3(m
2
r
2
) r
2
m
2
]
since we know that

r
1
= r
2
m
2
=m
1
=m
1
z
thus we can write the above expression as
B
T
=

0
3
4r
3
1
[(m
1
r
1
) r
1
( m
1
r
2
) r
2
]
or we can also write this as
B
T
=

0
3m
1
4r
3
1
[( z r
1
) r
1
( z r
2
) r
2
]
139
we can see from the diagram that
m
1
r
1
= sin r cos z m
1
r
2
= sin r +cos z
or from the diagram on the right we also nd
z r
1
=cos z r
2
= cos
thus the total magnetic eld can now be written as
B
T
=
3
0
m
1
4r
3
1
[cos( r
1
+ r
2
)]
but we know that
r
1
+ r
2
= 2a = 2sin r r
1
=
r
2
+h
2
and since we can see from the diagram that
cos =
h
r
1
=
h
r
2
+h
2
sin =
r
r
1
=
r
r
2
+h
2
using this we now nd that the total magnetic eld is
B
T
=
3
0
hrm
2(r
2
+h
2
)
5/2
r
To nd out what the surface current is we use
B =
0
(K z)
in order to solve for K we must take the cross products of both sides give respect to z we nd
z B = z (K z)
using the following relationship
A(BC) = B(A C) C(A B)
we nd
z B =
0
[K( z z) z( z K)
we know that
z K = 0 z z = 1
because they are orthogonal. We know that
z B =
r

z
0 0 1
3
0
hrm
2(r
2
+h
2
)
5/2
0 0
=
3
0
hrm
2(r
2
+h
2
)
5/2

thus we nd that the surface current is given by

K =
3hrm
2(r
2
+h
2
)
5/2

Problem # 9
140
Figure 1: A parallel plate capacitor is connected on one side to a voltage source that increases the voltage
across the capacitor linearly in time.
Figure 1 shows a parallel plate capacitor that is connected at one end (x = 0) to a source whose voltage
increases slowly and linearly with time as dV/dt = k. Assume that edge effects are negligible in this
problem, i.e. that a, b s .
a) Find the surface current K in the top capacitor plate (positive voltage) as a function of position x
along the capacitor.
Since we know that
Q(t) =CV
then
dQ(t)
dt
=C
dV
dt
=Ck
thus
I =Ck
the capacitance from a parallel plate capacitor is given as
C =
A
0
s
thus the current is given as
I =
xb
0
k
s
In order to satisfy the boundary conditions we must write this as
I =
(ax)b
0
k
s
and we know that the surface current is dened as
K =
I
b
x =
k
0
(ax)
s
x
b) Find the magnetic eld B inside the capacitor by two different ways: using the integral Am-
pere/Faraday law for two different loops such that one has a ux only of real current and the other only of
displacement current. From your expression for B (or by other means) determine the vector potential A
inside the capacitor.
We know that Maxwells equation is given by
B =
0
J+
0
0
E
t
141
and the integral Ampere/Faraday law can be written as
B =
I
B dl =
Z
A
0
J da+
0
0
Z
A
E
t
da
the Amperian loops used are given in the following diagram
For nding the magnetic eld using the real current we can use the diagram on the left, we can see that
since the loop is parallel with
E
t
= 0
Thus we have
I
B dl =
Z
A
0
J da =
0
I
enc
=
0
K(x)l
thus we nd that the magnetic eld is given by
B =

0
0
k
s
(ax) y B = 0 outside of the plate no z component)
now to nd this using the displacement current we know that
J = 0 no surface current enclosed
thus we nd
I
B dl =
0
0
Z
A
E
t
da =
0
k
s
(x
1
x
2
)(y
1
y
2
)
where we can also write the left hand side as
B(x
1
) B(x
2
) =

0
0
k
s
(x
1
x
2
) y
where this can also be written as
B =
0
k
s
(ax) y x
1
= x x
2
a
which is the same thing we found from the real current, which it should be. To nd what the vector
potential is we can use
B =A =
x y z
z
A
x
A
y
A
z

x
A
z

z
A
x
y =

0
0
k
s
(ax) y
Thus we can see that a suitable solution of the vector potential could be given as
A
z
=
0
k
s
(ax
1
2
x
2
) z A
x
= 0
142
Thus the magnetic eld inside the plate and the vector potential can be written as
B =
0
k
s
(ax) y A =
0
k
s
(ax
1
2
x
2
) z
c) Find B and A outside the capacitor.
For part be we concluded that
B = 0
for the vector potential outside we can guess a solution and see if it satises the boundary conditions we
can use
A
above
=
0
k
s
(az zx) x
0
k
s
(ax
1
2
x
2
+
1
2
z
2
) z
does this satisfy our boundary conditions?
A = 0
we can see that this is satised,
A = 0
this is satised if we assume that we are using the Coulomb gauge which is dened as this. What about
A
above
(z = 0) = A
below
(z = 0)
we can see that this is also satised. And nally
A
above
(z = 0)
n

A
below
(z = 0)
n
=
0
K
and we can see that this will also be satised by our vector potential. Thus we can conclude that
B
above
= 0 A
above
=
0
k
s
(az ax) x
0
k
s
(ax
1
2
x
2
+
1
2
z
2
) z
Problem # 10 Magnetic pressure on a solenoid
Consider a long solenoid that produces a magnetic eld of magnitude B in its interior.
a) Considering the magnetic energy per unit length in a solenoid of radius R and R+dR, show that
there is an outward pressure on the surface of the solenoid (i.e. on the many turns of wire that resemble a
continuous sheet of current) of magnitude P = B
2
/(2
0
).
We know that the energy stored due to a magnetic eld is
W =
1
2
0
Z
B
2
d =
B
2
2
0
V
where the volume of this solenoid is given by
dV = h[(R+dR)
2
R
2
]
thus the energy per unit length is given by
dW
h
=
B
2
2
0
[(R+dR)
2
R
2
]
143
We also know that the energy is related to the force by
dW =
Z
F dl = P
Z
Adr = 2hP
Z
R+dR
R
rdr =hP[(R+dR)
2
R
2
]
and we can see that this can also be written as
dW
h
= P[(R+dR)
2
R
2
] =
B
2
2
0
[(R+dR)
2
R
2
]
thus we nd that the pressure is given by
P =
B
2
2
0
b) Suppose a public utility proposes to moderate the ebb and ow of intermittent power sources by
storing energy in a large inductor, reaching a maximum eld of, say, 14 Tesla. What is the storage capacity
of this inductor, in kilowatt-hours / (meter)
3
? What would be the magnetic pressure on this solenoid in
atmospheres?
We know that
W =
B
2
2
0
V
W
V
=
B
2
2
0
and since we have been given the magnetic eld and we know what
0
is we nd
W
V
= 7.810
7
J
m
3
= 21.7
kWh
m
3
= 770 atm
c) Alternately, suppose the energy is stored instead in a large capacitor with, say, a dielectric constant
/
0
= 3 and a maximum allowable eld of 1.5 10
8
V/m. This storage would not require, for instance,
cryogenic cooling of a superconductor to avoid resistive heating of a current-carrying solenoid. What is
the storage capacity of this capacitor, also in kilowatt- hours / (meter)
3
? For reference, the energy content
of gasoline is about 100 kilowatt-hours / (meter)
3
.
We know that the energy stored in a capacitor with a dialectric media is given as
W =

0
2
Z

r
E
2
d =

0
r
2
E
2
V
thus
W
V
=

0
r
E
2
2
We know that
=
r
= 3
thus
W
V
=
3
2
0
E
2
given that we know what E is we nd
W
V
= 310
5
J
m
3
= .083
kWh
m
3
144
Chapter 8
Conservation Laws
Problem # 1
a) Consider two equal point charges q, seperated by a distance 2a. Construct the plane equidistant
from the two charges. By integrating Maxwells stress tensor over this plane, determine the force of one
charge on the other.
We can see that this conguration will cause a repulsive force on the top charge due to the bottom
charge. If we pick a point located a distance r from the origin a distance r
from the charge and at z = 0

we see that the electric eld here only has a y component (the vertical components cancel by symmetry).
E
x
= E
z
= 0 E
y
(r
) = 2Esin
Maxwells stress tensor is given as
T
i j
E
i
E
j
1
2
i j
E
2
+
1
B
i
B
j
1
2
B
2
(8.1)
145
The indices i and j refer to the coordinates x, y, and z, so the stress tensor has a total of nine components.
We can write the stress tensor explicitily in matrix notation as
T =
T
xx
T
xy
T
xz
T
yx
T
yy
T
yz
T
zx
T
zy
T
zz
One can form the dot product with a vectro a:

(a
T)
j
=

i=x,y,z
a
i
T
i j
If the force per unit volume is given as
f =
T
0
0
S
t
then the total force is given by
F =
I

T da
0
0
d
dt
Z
V
Sd (8.2)
where we used the divergence theorem to convert the rst term to a surface integral. In static situations the
second term drops out and we are left with
F =
I

T da (8.3)
We can see from our gure that if we were to use the xy plane as our boundary surface then
da = rdrd( z)
We can write Maxwells stress tensor as
T
xx
=
1
2
0
(E
2
x
E
2
y
E
2
z
) T
yy
=
1
2
0
(E
2
y
E
2
z
E
2
x
) T
zz
=
1
2
0
(E
2
z
E
2
x
E
2
y
) (8.4)
Since we know the net force is in the z dircetion it sufces to calculate
(
T da)
z
= T
zx
da
x
+T
zy
da
y
+T
zz
da
z
since we know that
da
x
= da
y
= 0
this becomes
(
T da)
z
= T
zz
da
z
(8.5)
We can see that
T
zz
=
0
2
E
2
y
E
z
= E
x
= 0 da
z
= rdrd
Thus the force is given Equation 3 and Equation 4 as
F =
0
2
I
E
2
y
rdrd
we nd that the electric eld at r and r
is given by
E
y
= 2Esin sin =
r
r
sin
2
=
r
2
r
2
146
The electric eld for a uniform charge is given by
E(r
) =
q
4
0
1
r
2
r
since we only have an electric eld in the y direction this becomes
E
y
=
2q
4
0
1
r
2
sin y
If we integrate over the entire xy plane and also a hemisphere extending out to innity, where E = 0 at
innity and thus the integral contributes nothing allows us to write the force as
F =
0
2
Z
2
0
d
Z

0
2q
4
0
1
r
2
sin
2
rdr
This simplies to
F =
q
2
4
0
Z

0
r
3
r
6
dr
From the diagram we can see that r
a
2
+r
2
thus
F =
q
2
4
0
Z

0
r
3
(a
2
+r
2
)
3
dr =
q
2
4
0
2r
2
+a
2
4(r
2
+a
2
)
2
0
Evaluating this integral gives
F =
q
2
16
0
a
2
z
Which is exactly what we would expect using
F =
q
2
4
0
r
2
z r = 2a F =
q
2
16
0
a
2
z
b) Do the same for charges that are opposite in sign.
147
We can see that for this conguration we nd the net force is attractive and if we consider the top
charge to be q than it will experience a net force in the z direction. We can see that the electric eld at
r and r
is now given as
E
x
= E
y
= 0 E
z
= 2Ecos
since we know that the net force is still in the z direction then we can choose
da = rdrd( z)
and thus we it works to solve for
(
T da)
z
= T
zz
da
z
where
T
zz
=

0
2
E
2
z
da
z
= rdrd
and so we nd the force to be
F =

0
2
Z
2
0
d
Z

0
2q
4
0
1
r
2
cos
2
rdr
where
cos =
a
r
cos
2
=
a
2
r
2
this now gives
F =
q
2
a
2
4
0
Z

0
r
(a
2
+r
2
)
3
dr
doing a u substitutin gives
u = a
2
+r
2
rdr =
du
2
gives
F =
q
2
a
2
8
0
Z

a
2
u
3
du =
q
2
16
0
a
2
z
which is exactly what is expected.
Problem # 2
Consider an innite parallel-plate capacitor, with the lower plate (at z = d/2) carrying the charge
density , and the upper plate (at z = +d/2) carrying the charge density +.
148
a) Determine all nine elements of the stress tensor, in the region between the plates. Display your
answer as a 33 matrix:

T
xx
T
xy
T
xz
T
yx
T
yy
T
yz
T
zx
T
zy
T
zz
since the Maxwell stress equation is given by

T
i j
E
i
E
j
1
2
i j
E
2
+
1
B
i
B
i
1
2
B
2
and assuming that the magnetic eld inside the capacitor is B = 0 we can write the nine components of
the stress tensor as
T =
0
E
x
E
x
1
2
xx
E
2
E
x
E
y
E
x
E
z
E
y
E
x
E
y
E
y
1
2
yy
E
2
E
y
E
z
E
z
E
x
E
z
E
y
E
z
E
z
1
2
zz
E
2
this can be written as
T =
0
1
2
E
2
x
E
2
y
E
2
z
E
x
E
y
E
x
E
z
E
y
E
x
1
2
E
2
y
E
2
z
E
2
x
E
y
E
z
E
z
E
x
E
z
E
y
1
2
E
2
z
E
2
x
E
2
y
Looking at the gure above we can conclude that

E
x
= E
y
= 0 E
z
=
0
z
The eld in this conguration is pointing in the - z direction this gives us
T =

2
2
0
1 0 0
0 1 0
0 0 1
b) Use Equation 8.22

F =
I
S
T da
0
0
d
dt
Z
V
Sd
to determine the force per unit area on the top plate. Compare Eq 2.51
Given the static situation
S = 0 da = z
we nd that the force is given by
f =
F
da
=

2
2
0
1 0 0
0 1 0
0 0 1
n
since we know the force is only in the - z direction we nd
f =
T =
2
2
0
z
149
which is identical to Equation 2.51.
c) What is the momentum per unit area, per unit time, crossing the xy plane (or any other plane parallel
to that one, between the plates).
According to Newtons second law. the force on an object is equal to the rate of change of its momen-
tum
F =
dp
mech
dt
where we can now write Equation 2 as
dp
mech
dt
=
I
S
T da
0
0
d
dt
Z
V
Sd (8.6)
The second integral represents momentum stored in the electromagnetic f ields themselves:
p
mech
=
0
0
Z
V
Sd
while the rst integral is the momentum per unit time f lowing through the sur f ace. Equation 6 is the
general statement of conservation o f momentum in electrodynamics. If we let P
mech
be the density of
mechanical momentum, and P
em
the density of momentum in the elds
P
em
=
0
0
S
then equation 6 in differential form says
t
(P
mech
+P
em
) =
T
and so we call -
T the momentum ux density. We have found this to be

P
ux density
=
T =f =

2
2
0
z
d). At the plates this momentum is aborbed, and the plate recoils (unless there is some non-electrical
force holding them in position). Find the recoil force per unit area on the top plate, and compare your
answer to (b). [Note : This is not an additional force, but rather an alternative way of calculating the same
force- in (b) we got it from the force law, and in (d) we get it by conservation of momentum.]
The recoil force on the top plate should be equal and opposite to the force exerted on the plate
F
recoil
a
=
dp
mech
dadt
=P
ux density
=
2
2
0
z
which is the same as (b), like Grifths implied in the problem. Same problem just two different ways of
looking at it.
Problem # 3
150
A charged parallel-plate capacitor (with uniform electric eld E = E z) is placed in a uniform magnetic
eld B = B x as shown by the gure
a) Find the electromagnetic momentum in the space between the plates.
The electromagnetic momentum is dened as
p
em
=
0
0
Z
V
Sd =
0
Z
(EB)d
thus we nd
p
em
=
0
EBsinV y
and since = /2 we nd that the momentum is given as
p
em
=
0
EBdA y
where the volume is given by
V = dA
b) Now a resistive wire is connected between the plates, along the z axis, so that the capacitor slowely
discharges. The current through the wire will experience a magnetic force; what is the total impulse
delivered to the system, during the discharge?
We know that the impulse is given as
I =
Z
Fdt
The magnetic force on a segment of current-carrying wire is given by
F
mag
=
Z
I(dl B)
151
thus the impulse can now be written as
I =
Z

0
Z
l
0
I(dl B)dt =
Z

0
I(l B)dt
since we know that l = d z and B = B x gives us
I =
Z

0
IdB( z x)dt =|dB yIt|
0
=dB(q() q(0) y = q(0)dB y
where q(t) = It and q() = 0, we assume that there is no charge left at t = . We also know that Gausss
law gives
Z
E da =
q
enc
0
EA
0
= q
enc
Thus we nd that the impulse is given by
I =
0
EBdA y
which is the same thing as
I =
Z
Fdt =
Z
dp
em
dt
dt = p
em
c) Instead of turning off the electric eld (as in (b)), suppose we slowely reduce the magnetic eld.
This will induce a Faraday electric eld, which in turn exerts a force on the plates. Show that the total
impulse is (again) equal to the momentum originally stored in the elds.
Since we know that Faradays law in integral form is given as
I
E dl =
Z
B
t
da
152
and in differential form
E =
B
t
we know that a changing magnetic eld in the x direction induces a Faraday electric eld in the y direction.
Using the Amperean loop shown in the gure we nd
I
E dl =
Z
b
a
E dl +
Z
d
0
E dl
Z
b
a
E dl
Z
d
0
E dl
we can see that the second and last term cancel and since the electric eld is in the y direction we nd
(E(0) +E(d))l =
t
(B() B(0))dl =
B(0)dl
t
y
and the force is simply
F = qE = q(E(0) +E(d)) = qd
B(0)
t
y
using this in the impulse equation yields
I =
Z
Fdt = qd
Z
B(0)
t
dt y = qdB y
but we have shown that
q =
0
AE
thus the impulse is
I =
0
EBdA y
which is the same thing as in the momentum stored in the elds.
Problem # 4
Imagine an iron sphere of radius R that carries a charge Q and a uniform magnetization M = M z. The
sphere is initially at rest.
a) Compute the angular momentum stored in the electromagnetic elds.
Since we know that the magnetic eld outside of a uniformly magnetized sphere is given by the mag-
netic dipole equation
B
dip
=

0
m
4r
3
[2cos r +sin
]
and we know that the electric eld inside this conducter is zero so we are only concerned with the electric
eld outside of this sphere which can be modeled as the electric eld of a point charge at a large distance
away
E =
q
4
0
r
2
r
the angular momentum density is given as
em
= r P
em
=
0
[r (EB)]
where we have dened the momentum density
P
em
=
0
0
S =
0
(EB)
153
thus we nd
EB =

0
qm
16
2
0
r
5
sin
the total angular momentum stored in the elds is given by

L
em
=
em
V
where
em
=

0
qm
16
2
r
4
sin
we know that

in cartesian coordinates is

thus the angular momentum density is
em
=

0
qm
16
2
r
4
sin(coscos x+cossin ysin z)
and the total angular momentum is
L
em
=
Z

em
d =

0
qm
16
2
Z

R
1
r
2
dr
Z
2
0
d
Z

0
sin
3
d z
which becomes
L
em
=
4(2)
0
qm
(3)16
2
R
and if
m =
4
3
R
3
M
gives
L
em
=
2
9
0
MqR
2
z
b) Suppose the sphere is gradually (and uniformly) demagnetized (perhaps by heating it up past the
Curie point). Use Faradays law to determine the induced electric eld, nd the torque this eld exerts,
and calculate the total angualr momentum imparted to the sphere in the course of the demagnetization.
154
If we let
dl =2s
da =s
2
z
thus we can use Faradays law
I
E dl =
Z

B
t
da
thus the Faraday electric eld is given as
E
far
=
dB
dt
s
2
2
=
1
3
0
dM
dt
R
2
sin
but we know that

s = Rsin
dB
dt
=
2
3
0
dM
dt
the torque is given as
N = r F
where the force is from the induced Faraday electric eld
F = qE
far
dF = dqE
far
=daE
far
and the total torque is given by
Z
N
0
dN =
Z
r dF =
Z
RdF( r
) =
Z
RdF(
) =
Z
RdF sin z
if we now dene
da = R
2
sindd
thus
N =
1
3
0
3
R
4
dM
dt

Z
d
Z

0
sin
3
d z =
1
3
24R
2
3
dM
dt
R
2
0
z
thus the total torque is given by
N =
2
9
0
qR
2
dM
dt
z
and thus the total angular momentum is given as
L
mech
= Ndt =
2
9
q
0
R
2
M z
c) Suppose instead of demagnetizing the sphere we discharge it, by connecting a grounding wire to
the north pole. assume the current ows over the surface in such a way that the charge density remains
uniform. Use the Lorentz force law to determine the torque on the sphere, and calculate the total angular
momentum imparted to the sphere in the course of the discharge. (The magnetic eld is discontinuous at
the surface... does this matter?)
We know that the force will now be given by
F = q(vB) = (KB)da
and we know that the torque is now given by
N = r F = (r (KB))da
155
using the BAC-CAB rule for the cross product of a cross product we nd
N = (K(r B) B(r K))da
we know that r K = 0 due to orthogonality, thus we nd
N = K(r B)da
we know that the magnetic eld will just be the average of the internal eld and the external eld
B =
1
2
0
m
2r
3
(cos r sin

) +

0
mcosR
3
2r
3
r +

0
msinR
3
4r
3

and
r B =

0
m
2r
3
cos
now we need to nd what the surface current is, we know that
q =
Z
da =
Z
2
0
d
Z

R
2
sind =2R
2
(1+cos)
and
dq
dt
=
d
dt
[2R
2
(1+cos)]
we also know that
=
Q
4R
2
thus we nd the current to be given by
I =
dQ
dt
1+cos
2
but we know that the surface current is given by

K =
I
l
but
l
= 2Rsin I =
dQ
dt
1+cos
2
thus the surface current is

K =
dQ
dt
1+cos
4Rsin
thus our total torque is given by

N =
dQ
dt
Z
1+cos
4Rsin
0
m
2
cos
da
but we know that

da = R
2
sindd

(sin z)
thus this integral be
N =
m
0
8R
2
dQ
dt
Z
2
0
d
Z

0
(sincos+sincos
2
)d z
156
since we know that
m
4R
3
=
M
3
thus we nd
N =

0
M
6
R
2
dQ
dt
Z
2
0
d
Z

0
(sincos+sincos
2
)d z
which gives
N =
1
3
0
R
2
M
dQ
dt
Z

0
(sincos+sincos
2
)d
which becomes
N =
2
9
0
MR
2
dQ
dt
and thus the total angular momentum is
L
mech
= Ndt =
2
9
0
MR
2
Q z
157
Chapter 9
Electromagnetic Waves
Problem # 1
Suppose
E(r, , , t) = A
sin
r
[cos(kr t) (1/kr)sin(kr t)]
, with

k
= c
(This is, incidentally, the simplest possible spherical wave. For notational convenience, let (kr t) u
in your calculations.)
a) Show that E obeys all four of Maxwells equations, in vacuum, and nd the associated magnetic
elds.
In regions of space where there is no charge or current, Maxwells equations read
(i) E = 0 (iii) E =
B
t
(ii) B = 0 (iv) B =
0
0
E
t
for (i) we nd
E = A
sin
r
[cosu(1/kr)sinu]
in spherical coordinates this gives

E =
A
r
2
[cosu(1/kr)sinu] = 0
to nd the magnetic eld we must use
E =
B
t
where letting
v
= A
sin
r
[cosu(1/kr)sinu]
E =
1
r sin
(sinv
r
1
r
r
(rv
158
which is given by
E =
A2sincos
r
2
sin
[cosu
1
kr
sinu] r
Asin
r
d
dr
[cosu
1
kr
sinu]
E =
A2cos
r
2
[cosu
1
kr
sinu] r
Asin
r
ksinu+
1
kr
2
sinukcosu
we know that this is equal to

A2cos
r
2
[cosu
1
kr
sinu] r
Asin
r
ksinu+
1
kr
2
sinukcosu
=
B
t
thus the magnetic eld is given by
B =
Z
A2cos
r
2
[cosu
1
kr
sinu] rdt
Z
Asin
r
ksinu+
1
kr
2
sinu
1
r
cosu
dt
solving this integral gives
B =
2Acos
r
2
sinu+
1
kr
cosu
r +
Asin
r
kcosu+
1
kr
2
cosu+
1
r
sinu
now for the divergence of the magnetic eld

B =
1
r
2
r
2
2Acos
r
2
sinu+
1
kr
cosu
+
1
r sin
Asin
2
kcosu+
1
kr
2
cosu+
1
r
sinu
which is simply 0
B = 0
now to for the last of Maxwells equation we use
B =
0
0
E
t
if we call
v
r
=
2Acos
r
2
sinu+
1
kr
cosu
=
Asin
r
kcosu+
1
kr
2
cosu+
1
r
sinu
Thus the cross product gives

B =
1
r

r
(rv
)
v
r
r
(rv
) =
Asin
k
2
sinu
1
r
2
sinu
2
kr
3
cosu
1
r
2
sinu+
k
r
cosu
and
v
r
=
2A
r
2
sinu+
1
kr
cosu
(sin)k
159
thus summing up these two contributions yields
B =
Asin
cr
ksinu+
1
r
cosu
=
0
0
E
t
now we know that
0
E
t
=
1
c
2
A
sin
r
[cosu(1/kr)sinu
=
1
c
2
Asin
r
k
ksinu+

k
1
r
cosu
=
Asin
cr
ksinu+
1
r
cosu
thus we can see that both sides are equal.

b) Calculate the Poynting vector. Average S over a full cycle to get the intensity vector I. (Does it point
in the expected direction? does it fall like r
2
, as it should?)
We know that the Poynting vector is given by
S =
1
0
(EB)
given the magnetic eld found in part (a) and using the electric eld given gives (minus a lot of algebra)
S=
1
0 0 A
sin
r
[cos(kr t) (1/kr)sin(kr t)]
2Acos
r
2
cosu+
1
kr
sinu
Asin
r
kcosu+
1
kr
2
cosu+
1
r
sinu
which will give

S =
A
2
sin
2
r
2
[cosu(1/kr)sinu]
kcos u+
1
kr
2
cosu+
1
r
sinu
r
+
2A
2
sincos
r
3
[cosu(1/kr)sinu]
cosu+
1
kr
sinu
Simplifying the r component gives

S( r) =
A
2
sin
2
0
r
2
2
r

1
k
2
r
3
sin2u
2
+
1
kr
2
cos2ukcos
2
u
r
we know that the intensity is given by
I =S
sinucosu =sin2u =cos2u = 0 cos
2
u =sin
2
u =
1
2
we nd that the intensity will now be given by
I =S =
A
2
ksin
2
2
0
r
2
r
c) Integrate I da over s spherical surface to determine the total power radiated [Answer : 4A
2
/3
0
c]
160
we know that
P =
I
I da =
A
2
k
2
0
Z
2
0
d
Z

0
sin
3
=
4
3
A
2
2k
2
0
=
4A
2
3
0
c
Problem # 2
Find the elds, and charge and current distributions, corrsponding to
V(r, t) = 0, A(r, t) =
1
4
0
qt
r
2
r
We know that
E =V
A
t
=
q
4
0
r
2
r
and the magnetic eld is given by
B =A = 0
to nd the charge distribution we can take the divergence of the electric eld
=
0
E =
q
4
r
r
2
= q
3
(r)
and nally the current density
J =
1
0
B
0
E
t
= 0
since we know that the magnetic eld is 0 and also that the electric eld is constant in time.
Problem # 3
Use the gauge function = (1/4
0
)(qt/r) to transform the potentials in Prob. 10.3 and comment
on the result.
Using Equation 10.7 we nd
A
= A+ = 0 V
=V
t
thus
B = A
= 0
we can also see that
E =V
t
=
q
4
0
r
2
r
and the current distribution is again
J = 0
These are the same conclusions, this follows from the fact that we can add any to A, provided we
simultaneously subtract /t from V.
Problem # 4
Find all the elements of the Maxwell stress tensor for a monochromatic plane wave traveling in the z
direction and linearly polarized in the x direction
E(z, t) =

E
0
e
i(kxt)
x

B(z, t) =
1
c
E
0
e
i(kxt)
y
161
or taking the real part
E(z, t) = E
0
cos(kz t +) x B(z, t) =
1
c
E
0
cos(kz t +) y
Does your answer make sense? (Remember that

T represents the momentum ux densitu.) How is the
momentum ux density related to the energy density in this case?
T
i j
E
i
E
j
1
2
i j
E
2
+
1
B
i
B
j
1
2
i j
B
2
(9.1)
The indices i and j refer to the coordinates x, y, and z, so the stress tensor has a total of nine components.
We can write the stress tensor explicitily in matrix notation as
T =
T
xx
T
xy
T
xz
T
yx
T
yy
T
yz
T
zx
T
zy
T
zz
We can write Maxwells stress tensor as

T
xx
=
1
2
0
(E
2
x
E
2
y
E
2
z
) +
1
2
0
(B
2
x
B
2
y
B
2
z
) (9.2)
T
yy
=
1
2
0
(E
2
y
E
2
z
E
2
x
) +
1
2
0
(B
2
y
B
2
z
B
2
x
) (9.3)
T
zz
=
1
2
0
(E
2
z
E
2
x
E
2
y
) +
1
2
0
(B
2
z
B
2
x
B
2
y
) (9.4)
and
T
xy
= T
xz
= T
yx
= T
yz
= T
zx
= T
zy
= 0
and since we know
B
x
= B
z
= E
y
= E
z
= 0
T
xx
=
1
2
0
E
2
x
1
2
0
B
2
y
(9.5)
T
yy
=
1
2
0
E
2
x
+
1
2
0
B
2
y
(9.6)
T
zz
=
1
2
0
E
2
x
1
2
0
B
2
y
(9.7)
thus Maxwells stress tensor in matrix notaion is given by
T =
1
2
0
E
2
x
1
2
0
B
2
y
0 0
0
1
2
0
E
2
x
+
1
2
0
B
2
y
0
0 0
1
2
0
E
2
x
1
2
0
B
2
y
where the off diagonal elements are all zero. Plugging in our expressions for the electric eld and magnetic
eld gives
T =
0 0 0
0 0 0
0 0
0
E
2
0
cos
2
(kz t +)
162
this makes sense because the wavefront is moving in the z direction. In this case the momentum density is
equal to the negative of the momentum density. Grifths denes the energy density on Equation 9.55 as
u =
0
E
2
0
cos
2
(kz t +)
which is
u =
T =T
zz
=
0
E
2
0
cos
2
(kz t +)
Problem # 5
a) Suppose you imbedded some free charge in a piece of glass. About how long would it take for the
charge to ow to the surface?
The equation that governs how charge dissipates through a conductor is given by Equation 9.120 as
f
= e
(/)t
f
(0)
thus any initial free charge densiy
f
(0) dissipates in a characteristics time

where
1
glass
=
glass
(10
10
10
14
)
where
glass
is the resitivity and
glass
is the conductivity and the permitivity is
glass
4.97.5
0
we nd that the characteristic time is given by
0.433 sec
1
glass
10
10
glass
4.9
0
6198 sec
1
glass
10
14
glass
7.0
0
b) Silver is an excellent conductor but its expensive. suppose you were designing a microwave exper-
iment to operate at a frequency of 10
10
Hz. How thick would you make the silver coatings?
We know that the distance it takes to reduce the amplitude of a wave by a factor of 1/e (about a third)
is called the skin depth which is dened as
d
1
(9.8)
where is dened as

1+
2
1
1/2
(9.9)
thus if the thickness of our silver sheet is given by 8 we will have a coating with a thicknes that will stop
the microwaves by at least 1/3. Given that
silver
0
1
silver
= 1.5910
8
silver
=
0
163
since we also know that = 2f where f is the frequency we nd
= 1.5710
6
m
1
thus we should make this
d 6.3610
7
m
we should probably make it a little thicker than this to stop all EM waves, i.e
width 2d 1.2710
6
m
c) Find the wavelength and propgation speed in copper for radio waves at 1 Mhz. Compare the corre-
sponding values in air (or vacuum).
We know that the wavelength and speed of propgation is given by
=
2
k
v =

k
where k is dened as
k
1+
2
+1
1/2
for copper
1
copper
=
copper
1.6810
8
= 2f = 210
6
s
1
thus
copper
5.910
7
we also know that
copper
(220)
0

we can know that
2
1
1+
2
1
and thus we can write
k
2
where
0
we nd
k
0
2
1.510
4
m
1
thus the wavelength is
=
2
k
4.1110
4
m
and the velocity is given as
v =

k
419 m/s
1
Problem # 6
A plane wave travels in the z direction in a conductor with real conductivity.
164
a) Find the instantaneous and time-averaged power loss per unit volume due to resistive heating for
any z.
The electric eld and magnetic elds in this case are given by Equation 9.138 in Grifths as
E(z, t) = E
0
e
z
cos(kz t +
E
) x (9.10)
B(z, t) = B
0
e
z
cos(kz t +
E
+) y
The power loss due to resistive heating is given as
dP = d(VI)
we know that
dV = Edx
and
dI = J da = Jdydz
thus
dP = JEdxdydz
where
J =E
thus the power loss per unit volume is given by
P
V
=E
2
= E
2
0
e
2z
cos
2
(kz t +
e
)
and the time average power loss is given as
P
V
=

2
e
2z
E
2
because we know that
cos
2
(kz t +
E
) =
1
2
b) Find the total power loss per unit area between z = 0and z .
To nd the total power loss per unit area we must integrate the power per volume
P
A
=
Z

0
P
V
dz =
1
2
E
2
0
Z

0
e
2z
dz =

4
E
2
0
e
2z
c) Find the time-averaged Poynting vector at any z.
The Poynting vector is dened as
S =
1
(EB)
using Equation 10 and Equation 11 we nd
S =
B
0
E
0
e
2z
[cos(kz t +
E
)cos(kz t +
E
+)] z
165
if we let x = kz t +
E
and =/4 we nd
S =
B
0
E
0
e
2z
[cos(x)cos(x +)] z =
B
0
E
0
e
2z
[cos
2
xcoscosxsinxsin] z
The intensity is given as
I =S =
B
0
E
0
2
e
2z
cos z =
E
2
0
2
e
2z
Kcos z
since we know that
tan =

k
cos =
k
k
2
+
2
and since we know that
K =
k
2
+
2
we nd
Kcos = k
thus the intensity is given as
I =S =
k
E
2
0
2
e
2z
z
d) Compare the value of your result for parts (b) and (c). Are the results in accord with one another?
Explain.
at z = 0 the magnitude of these two functions should be the same. For the Poynting Vector we nd
S(z = 0) =
kE
2
0
2
and for the energy per unit area per unti volume we nd
P(z = 0)
A
=

4
E
2
0
we need to show
k
=

2
this can be done using
k =
2
1+
2
+1
1/2
1+
2
1
1/2
=
2
thus we nd
k
=

2
which is what we were looking for all along thus these two quantities are equal at z = 0
Problem # 7 A parallel-plate capacitor at high frequencies
When one designs an electronics circuit with conventional devices like resistors, capacitors and in-
ductors, one is wise to keep in mind that everything changes at high frequencies. To illustrate this, let
166
us consider a capacitor composed of two closely spaced and parallel circular plates, with radius a and
separation h (h a).
a) Consider rst the quasi-static situation, by which we mean the low-frequency response of the
capacitor. If the capacitor is driven with a single-frequency sinusoidal voltage V =V
0
e
it
, what are the
electric and magnetic elds, E(r, t) and B(r,t), produced inside the capacitor? You can write these as
complex quantities, but keep in mind that the physical values of elds, etc., are given by the real parts of
the complex functions.
We know that
E(r, t) =
V
h
z =V
0
e
it
h
z =
V
0
h
(cost +i sint) z
thus the real part is given by
E(r, t) =
V
0
h
cost z
we also know from Maxwells equation
1
c
2
E
t
=B
and using Stokes theorem
1
c
2
Z
E da =
I
B dl
we know
da =r
2
dl = 2r
thus we have
r
2
c
2
dE
dt
= B2r
where
dE
dt
=
V
0
h
sint
thus we nd the magntic eld to be given as
B(r, t) =V
0
r
2hc
2
sint
b) From the formula relating the magnetic energy and self-inductance of an inductor and from your
answer above, what is the self-inductance for this capacitor? In determining this energy, you will have to
consider a specic volume over which to integrate the energy density - take this to be the inner volume of
the capacitor.
We know that the energy is given by
W =
1
2
LI
2
Equation 7.34 gives the energy of the magnetic eld as
W =
1
2
0
Z
all space
B
2
d
thus
1
2
0
Z
all space
B
2
d =
1
2
LI
2
167
the left hand side becomes
1
2
0
Z

V
0
r
2hc
2
sint
2
d d = 2rhdr
which gives
1
2
0
Z

V
0
r
2hc
2
sint
2
2hrdr =
h
V
0
2hc
2
2
sin
2
t
Z
a
0
r
3
dr
thus
W =
a
4
4
0
h
V
0
2c
2
2
sin
2
t
to nd the inductance
a
4
4
0
h
V
0
2c
2
2
sin
2
t =
1
2
LI
2
(9.11)
if we let
I =C
dV
dt
C =
a
2
0
h
dV
dt
=
V
0
h
sint
thus
I
2
=
a
2
0
h
V
0
h
sint
2
substituting this into Equation 15 gives a inductance of
L =

0
h
8
c) From the standard capacitance C and the inductance L that you just calculated, determine a resonant
frequency
0
for the capacitor as you would for a standard LC circuit. The frequency
0
is a good
benchmark for what qualies as low or quasi-static vs. high frequency operation of the capacitor.
We know that for n L-C circuit the angular frequency is dened as
=
1
LC
we know that
LC =
a
2
8c
2
thus the resonant frequency is given by
0
=
c
a
8
d) In (a), you will have determined the magnetic eld generated by the time-varying electric eld from
Faradays law. But, of course, the time-varying magnetic eld also generates an electric eld according to
the Ampere-Maxwell law. Taking the magnetic eld discussed in parts (a) and (b), calculate this correction
to the electric eld of a capacitor.
The Ampere-Maxwell law gives
E =
B
t
168
using Stokes theorem we nd
I
E dl =
d
dt
Z
B da
where
da = hdr
and using the magnetic eld from part a) we nd

E(0)hE(r)h =
d
dt
Z
r
0
V
0
r
2hc
2
sint(hdr)
where
d
dt
Z
r
0
V
0
r
2hc
2
sint(hdr) =

2
V
0
4c
2
r
2
cost
dening
E(0) = 0
we nd
E(r) =
2
V
0
4hc
2
r
2
cost z
this is the correction to the electric eld due to the changing magnetic eld. Thus the total electric eld is
E
tot
=
V
0
h

2
V
0
4hc
2
r
2
cost z
e) You should nd that, at this level of approximation, the electric eld inside the capacitor is parabolic
as a function of radius r. At what frequency does the electric eld at the outer edge of the capacitor a
go to zero? How does it compare with the frequency
0
determined above?
If we set the electric eld equal to zero we nd
V
0
h

2
V
0
4hc
2
r
2
= 0
we nd that the frequency that makes this go to zero is
=
2c
a
which is smaller by
2.
Beyond the approximation discussed above, a complete solution for the electric and magnetic elds
inside a capacitor can be found that is consistent with Maxwells equations. As often in problems with
cylindrical symmetry, the solutions for the spatial variations of E and B are Bessel functions. These
functions also serve to characterize the resonant modes inside a cylindrical metal can.
169

110A Formulas Solutions

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

110A Formulas Solutions

Uploaded by

Copyright:

Available Formats

Electricity and Magnetism: Physics 110A

Written by Roger L. Grifth @UCB fall 2007

and a unit vector in the direction from r

sincos sinsin cos

= coscos x+cossin ysin z

The curl of A is given by

thus we nd the components to be given by

A is not constant with the same argument as above except that r,

in cartesian coordinates the components of A are given by

sincos coscos sin

thus we nd the components to be given by

adding up these vectors yield

To nd the curl we use

thus we know that the scalar potential is

(b) Show that F

given that F(r) =

in our case r = thus we have

is the seperation vector from the location of q to Q, r is the magnitude and

thus the electric eld of a line charge is

for a surface charge

and for a volume charge

is contnuous, as for instance at the surface of a uniformly

thus the charge density is given by

The plot is given by

38 = 6.16 6.16 > 6.0

and so the capacitance is given by

where A is the area of the plate. A gure is given by

From the diagram we can also infer that

= n total number of charges

thus we know that

and since we are integrating over the plane this means =

simpyfying this term yields

and nally we can put these two solutions together to get

substituting the above expression for B

we can use our knowledge of Legendre polynomials to solve this problem

for inside the sphere we nd

knowing that the derivative of the term is

For this problem we can just start with

setting like terms equal to each other gives

and for the outside we have

. Now to solve the radial differential equation, which is a non-linear homo-

/s term because it is negligable.

we can expend this using Taylor expansion to nd

and since we are only concerned with Q

but we know that

using the relationship from problem 3.45

and since we are only looking for Q

= coscos x+cossin ysin z

and we also know that the total force is given by

now that we have the electric eld we can nd the force

we can also write this as

but since we know

and from condition 2 we nd

from the last equation into the previous one yields

plugging this into either equation yields

term as the one con-

knowing this allows us to nd the capacitance per unit length

is an element of length along

5.4 The Divergence and Curl of B

c with respect to (laboratory) frame S

, the electric eld is transformed as

solve for the motion

has velocity components