Sunteți pe pagina 1din 16

Chapter 28

Physics 1C 09-30-05
Fall 2005
P. Kraus
Reading Assignment: Giancoli Chap. 28
——————————————————————————————————————————

1 Lightning Review
1.1 Electrostatics
Let’s first recall the most basic aspects of electrostatics, so that we can then move on to
electromagnetism. A concise statement of electrostatics is that there is a force between any
pair of charges, given by Coulomb’s law
1 Q1 Q2
F~ = r̂ (1)
4π0 r2
The force is repulsive for same sign charges, and attractive for opposite sign charges. Electro-
magnetism obeys the principle of superposition, so to find the force on a given charge we just
add up all the contributions from all the other charges.
The above paragraph summarizes the entire physical content of electrostatics. Nonetheless,
certain mathematical developments are highly useful in solving practical problems. It is useful
to say that a charge produces an electric field given by

~ = 1 Q
E r̂ (2)
4π0 r2
and that in the presence of an electric field a charge q experiences the force

F~ = q E
~ (3)

Even better, when all charges are stationary (which is the definition of electrostatics) it can
be useful to work with the potential energy rather than the force. The electric potential V is
defined by saying that the energy of a charge q is given by U = qV . The potential due to a
single charge is
1 Q
V = (4)
4π0 r
The total potential at a given point is given by summing up the contributions from all charges.
Since potential energy is the line integral of the force, we can write
Z b
V (b) − V (a) = − E ~
~ · dl (5)
a

This formula establishes that the electric field and the electric potential contain equivalent
information.
If we draw a sphere around a certain charge q Coulomb’s law shows that

~ · r̂ = q
Z
E ~ = 4πr2 E
~ · dA (6)
0
which is independent of the sphere radius. In fact, we get the same result for the integral even
if the surface is not a sphere but some more general shape. This is Gauss’ law. It is physically

1
Figure 1:

equivalent to Coulomb’s law, but often more helpful in solving problems. Gauss’ law is easy
to understand in terms of field lines. We think of field lines emanating radially from a charge.
The density of the field lines then drops off as 1/r2 , which therefore encapsulates Coulomb’s
law. The electric flux through a surface is then proportional to the number of field lines passing
through the surface (weighted by plus or minus for ingoing or outgoing field lines). It is then
geometrically clear that the number of field lines passing through a surface surrounding the
charge can’t change when we deform the surface, and this is the content of Gauss’ law.

1.2 Right hand rule


When we encounter magnetism in the next subsection we make frequent use of cross products.
It is crucial to have a good grasp of the cross product. The right hand rule is a basic tool for
this.
Suppose we have two vectors A~ and B.
~ We then define another vector C ~ as the cross product
of the first two: C~ =A ~ × B.
~ The definition is that C ~ is perpendicular to both A~ and B ~ and
has a length given by
~ = |A|
|C| ~ B|
~ sin θ (7)
~ and B.
where θ is the angle between A ~ The above definition actually leaves a two-fold ambiguity
~
in the direction of C, and this is resolved by the right hand rule (Fig. 1 (Looking at Fig. 11-1
in the book will probably be clearer) To compute A ~×B ~ you start with your right hand in
“handshake position”. Extend your fingers in the direction of A.~ Then make sure your hand is
~
oriented such that you can curl your fingers so that they end up pointing in the direction of B.
~
Your extended thumb is then pointing in the direction of C. Make sure you understand
this! If confused, seek help from me or the TAs!

1.3 Magnetism
Electrostatics is concerned with stationary charges. Now we allow our charges to move. Mov-
ing charges are the origin of magnetic fields. A charged particle moving in a magnetic field
experiences a force which is perpendicular to the velocity of the particle and to the magnetic

2
field
F~ = q~v × B
~ (8)
This is the Lorentz force law, and we can take it to define what we mean by a magnetic field.
So in order to feel a magnetic charge a particle needs to be moving; similarly, one also needs
moving charges to produce magnetic fields. One might have expected there to be a simple
formula for the magnetic field of a single moving charge (analogous to Coulomb’s law), but this
is actually the wrong way to think. Instead, the simplest situation is to consider steady currents
(i.e. time independent), formed for example from a uniform distribution of charge moving at
constant velocity in a wire. The study of magnetism due to constant currents is magnetostatics.
Later we will consider currents which change in time, which is the subject of electrodynamics.
If we consider a steady current, (8) implies that the infinitesimal force on an infinitesimal
element of the current of length d~l is
~ ×B
dF~ = I dl ~ (9)

Of course the most familiar example of magnetism has to do with ferromagnets (the kind
on your refrigerator). Ironically, understanding how these works is far from trivial, and indeed
requires ideas outside of classical physics, namely quantum mechanics. So we won’t say too
much about ferromagnets in this course.

2 Chapter 28: Sources of Magnetic Field


2.1 Basics
As discussed above, magnetic fields arise from moving charges; i.e. from currents. We now
want to compute the precise magnetic field arising from a given current. This is somewhat
more subtle than computing the electric field due to a point charge. In that case the electric
field points out radially. But for currents, the magnetic field essentially circles around the
current.
Note that one might have expected there to exist magnetic charges: objects from which the
magnetic field would emanate out radially, in analogy with electric charges. In fact, there are
theoretical reasons to expect the existence of such “magnetic monopoles”. But they have never
been observed, and are at best extremely rare, so we need not concern ourselves with them
here. For us, magnetic fields come from currents.
As emphasized above, for now we will be considering currents which are constant in time –
steady currents. Current is defined as the charge moving past a given point per unit time, and
measured in amperes:
1 ampere = 1 coulomb/second (10)
It has to be stressed that a single point charge moving at constant velocity is not a steady
current, since at a given point in space it is there one moment and gone the next. Instead,
steady currents correspond to uniform distributions of charges all moving at constant speed.

2.2 Field of long straight wire


Let’s demonstrate the ability of currents to produce magnetic fields. First of all, we need a
way to measure magnetic fields. The simplest thing to use is basically a compass needle. For

3
Figure 2:

now, we don’t even need to know how this works, except to say that the needle points in the
direction of the magnetic field. Of course, a magnetic field also has a magnitude which is not
so easily measured by a compass, but let’s not worry about that for now.
Now, the simplest current setup is that of a long straight wire. Let’s measure the resulting
magnetic field. Demo: long straight wire. What we find is that the magnetic field circles
the wire (Fig. 2). The direction is given by the right hand rule (thumb in direction of current
yields fingers in direction of magnetic field). What is the magnitude of the field? We could in
principle use our compasses to measure this, by measuring the torque required to displace the
needle from its equilibrium position. If we did this we would find that the field is proportional
to the current carried by the wire. It is intuitively obvious that the field should drop to zero as
we get further from the wire. Careful measurement would show that it drops off as the inverse
power of the distance. The magnetic field outside a long straight wire is therefore proportional
to I/r, and we give the constant of proportionality the name µ0 /(2π):

µ0 I
B= , µ0 = 4π × 10−7 T · m/A (11)
2π r
For example, suppose we measure the field 1 cm away from a wire carrying 1 amp. We find
the field to be
2 × 10−7 T m · 1A
B= = 2 × 10−5 T (12)
10−2 mA
For comparison, the magnetic field produced by the earth is about 5 × 10−5 T as measured at
the earth’s surface.
We now know that a straight wire produces a magnetic field circling around it. We also
know that a moving charged particle in a magnetic field experiences the force (8). Since a
current is nothing but moving charged particles, we predict that two current carrying wires
should experience a force. The right hand rule shows that two parallel wires will attract if
the currents run in the same direction, and repel if they run in opposite directions. Demo:
parallel wires.
Let’s derive the formula for the force by combining (9) and (11) we find the force on wire 1
due to wire 2 to be
µ0 I1 I2
F1 = I1 l1 B2 = l1 (13)
2π r
Since the result is proportional to l1 the force is infinite for an infinite wire. However, the force
per unit length is finite, which is the physically relevant quantity. In particular, this means
that the acceleration of the wires towards each other is finite and nonzero.

4
Figure 3:

Figure 4:

2.3 Fields of other current distributions


Before developing the precise formulas let’s try to get an intuitive feel for the magnetic fields
produced by other steady current distributions. We can think of breaking up the current into
many pieces, consider the effect of each piece, and then add them up to get the total field.
That this is a valid procedure is equivalent to saying that the principle of superposition holds
in this context. At this intuitive level we just need to use that a little straight piece of current
gives a magnetic field which curls around and drops off with distance.
Let’s consider the field produced by a square current loop (Fig. 3) Very near each side we
can forget about the other sides, due to the 1/r behavior. In the middle of the loop the right
hand rule shows that the four sides reinforce each other. The field lines thus pass through the
loop, and then eventually curve back around outside the loop. The total number of field lines
passing through the inside of the loop is the same as the number outside. That is to say, each
field line is a closed loop without end or beginning. In fact, if the magnetic field lines did begin
or end somewhere, we would say that there is a magnetic monopole there. Instead, we have
what’s known as a magnetic dipole.
The same basic story holds for a circular current loop (Fig. 4). It is very useful to compare
this with the field produced by a ferromagnet (Fig. 5). We see the same basic thing. At a crude
level, you can imagine a ferromagnet as being composed of a bunch of little current loops1
1
The full story is a good deal more complicated.

5
Figure 5:

2.4 Biot-Savart Law


We now turn to the question of the magnetic field produced by an arbitrary steady current.
We take advantage of the superposition principle to first focus on the field produced by each
infinitesimal element of current, and then add up the results at the end. In particular, we
consider a wire with current I and focus on an infinitesimal length of wire d~l. What magnetic
field does it produce? Let’s try to guess the answer. Let’s consider the force between two such
current elements. We know from formula (27.4) or the Lorentz Force Law (8) that the force on
element 1, say, is
dF~1 = I1 d~l1 × B
~2 (14)
where B~ 2 is the field produced by element 2. Alternatively, we could have considered the force
on element 2 due to element 1
dF~2 = I2 d~l2 × B
~1 (15)
Now, the key point is that we must have F~1 = −F~2 . If the forces didn’t cancel out in this
way then a single closed current loop could experience an overall net force and accelerate itself
away in violation of momentum conservation. A moments thought shows right away that the
field produced has to be proportional to the current and to the length, otherwise the forces
wouldn’t cancel. To get a relative minus sign we’ll need the result to be proportional to r̂,
the unit vector extending from the element to the measured location of the field. Finally, the
statement that the field curls around the current means that B ~ should be orthogonal to both
~
dl and r̂. Altogether, we are led to
~ = f (r)Id~l × r̂
dB (16)

where f (r) is some function of the distance which is not determined by our analysis. It must
be taken from experiment, which exhibits a 1/r2 law, yielding the infinitesimal version of the
Biot-Savart law as
dB~ = µ0I d~l×r̂ (17)
4π r2
The total field of some arbitrary current distribution is then given by summing up (inte-
grating) the field due to all the current elements

~ =
Z
~ =
Z
µ0 I d~l × r̂
B dB (18)
4π r2

Example: field of long straight wire

6
Figure 6:

Consider an infinitely long wire extending up the y-axis (Fig. 6). Let’s find the field a
distance R from the wire. We may as well consider our point be at y = 0. Consider the
contribution from a current element at location y. We have
R
r= (19)
sin θ
Further
|d~l × r̂| = dy sin θ (20)
Also,
R 1 R
 
tan θ = ⇒ dy = Rd = − 2 dθ (21)
y tan θ sin θ
The direction of the field is given by the right hand rule, pointing into the paper at the indicated
point, and the magnitude of the field is

~ = µ0 I Z π sin θ µ0 I µ0 I
|B| dθ =− cos θ|π0 = (22)
4π 0 R 4πR 2πR
This is in agreement with our earlier claim.

Example: field of circular current loop


Consider a circular current loop of radius R. In this case one can “easily” write down an
integral for the magnetic field at an arbitrary location, but the integral can’t be done in closed
form. One expects simplifications at symmetrical positions, which in this case means on-axis.
So let’s find the field a distance x up the axis of the loop (See Fig. 7). Referring to the figure,
it’s clear that
|d~l × r̂| = dl (23)
The magnetic field has a component in the plane of the loop and a component perpendicular to
the plane of the loop. By symmetry, we can see that the former contributions cancel when we
integrate over the loop. The remaining component is proportional to cos φ. So the contribution
of a current element is
~ cos φ = µ0 Idl cos φ
|dB| (24)
4πr2
Since r and φ are the same for all points on the loop, the result for the full loop is given by
replacing dl by 2πR
~ = µ0 IR cos φ
|B| (25)
2r2
7
Figure 7:

Finally, we use trig to express everything in terms of x


R R
r2 = R2 + x2 , cos φ = =√ 2 (26)
r R + x2
giving
~ = µ0 IR2
|B| (27)
2(R2 + x2 )3/2
and the direction of the field is parallel to the axis.
~ =
The maximal value is attained at the center of the loop, x = 0, which gives |B| µ0 I
. In
2R
the other limit of large x the leading behavior is

~ ≈µ0 IR2
|B| , xR (28)
2x3
The 1/x3 behavior for large distance is to be compared with the inverse square law for the
electric field of a point charge. The distinction here is between a dipole in the magnetic case,
versus a monopole in the electric case. We have stated that magnetic monopoles have not been
observed; a discovery of such an object would be the same as finding an object whose magnetic
field falls off as an inverse square law2
It’s useful to phrase the result (28) in terms of the dipole moment of the loop. The dipole
moment is defined as
µ = IA = πR2 I (29)
so that (28) becomes
µ0 µ
~ ≈
|B| , xR (30)
2π x3
This is useful for the following reason. Note that I and R appear only through µ. Now suppose
we considered another current loop, not necessarily circular. We can again define the dipole
moment as above. For a complicated shaped loop the magnetic field produced will certainly
differ from the circular. However, it turns out that the leading behavior for large x always
takes the form (30). In fact, although we have only discussed the field on the axis, in fact the
leading long range field everywhere is expressed solely in terms of the dipole moment. This
is directly analogous to the statement that the leading long range electric field produced by
some arbitrary distribution of charges is given by (2) where Q is the total charge (the electric
monopole moment). The lesson here is that even for complicated setups the leading long-range
behavior of the electric and magnetic fields is quite simple.
2
Note that the field of an straight wire falls off very slowly, as 1/R. Our comments above only apply to the
case of a finite size object.

8
Figure 8:

2.5 Ampere’s law


We now want to discuss the magnetic analog of Gauss’ law (6). As with Gauss’ law, the
result will be extremely useful for solving problems with high symmetry, and is also of great
theoretical interest in terms of formulating electromagnetism in a concise fashion. Ampere’s
law can actually be thought of as a restatement of the Biot-Savart law (or vice versa).
To motivate Ampere’s law let’s return to the example of the infinite straight wire, with
field given by (11). Consider an imaginary circular loop of radius r, centered on the wire and
lying in a plane orthogonal to the wire. The magnetic field of the wire is tangent to this loop.
Therefore we can easily compute the following integral
Z
~ · d~l = 2πr|B|
B ~ = µ0 I (31)
loop

That’s interesting – the result is independent of the loop radius r due to a cancellation between
the field dropoff and the increase of the loop circumference.
Now consider a more general loop. Instead of going all the around the circle we go for an
angle δθ, then go either in or out radially, then go anotherR angle δθ0 around the circle, then
go radially, and so on (Fig. 8) Now what is the result for loop B ~ · d~l? Well, the radial paths
don’t contribute since they are perpendicular to the magnetic field. And as above, an angular
segment δθ contributes rδθ|B|~ = µ0 I δθ . Now, for a complete loop all the δθ’s have to add up

to 2π, so when we add up all the contributions we again get (31).
We can do more. We can further allow for loop segments which go parallel to the wire. Like
for the radial segments these are perpendicular to the field and so don’t change the conclusion.
It should be intuitively clear that if we make the individual segments very small then we
can approximate any continuous loop shape to arbitrary accuracy. This strongly suggest (and
it can be proven) that Z
B~ · d~l = µ0 I (32)
loop
for any loop that encloses the wire.
Next, what happens if we consider a loop that doesn’t enclose the wire? Let’s consider
the simplest case of a 4 sided loop with radial and angular segments (Fig. 9). The key point
is that the two angular segments cancel each other because if one is parallel to the magnetic
field then the other is anti-parallel. Since they span the same angle they give equal magnitude
contributions to the integral, but their opposite signs makes them cancel. The radial segments
don’t contribute as before. So we learn that
Z
~ · d~l = 0 if loop doesn0 t enclose wire
B (33)
loop

9
Figure 9:

Figure 10:

By the same logic as before, this conclusion holds for an arbitrary shaped loop.
We now have the general statement of Ampere’s law in the case of a straight wire
R
~ · d~l = µ0 Iencl
B (34)
loop

where the loop can have an arbitrary shape, and on the right hand side we just have the current
enclosed by the loop.
The next natural thing to look at is to replace the infinite straight wire by some more
general current distribution. The remarkable fact is that (34) continues to be valid. The
precise definition of Iencl is as follows. Think of a two-dimensional surface which “fills in” the
loop (like a soap bubble suspended in a wire frame.) This surface will be pierced by a certain
amount of current. A current contributes with a plus or minus sign according to the right hand
rule; that is, a plus sign arises if the current is in the direction of your thumb when your right
hand goes around the loop.

More on the derivation of Ampere’s law3


We claimed that the Biot-Savart law is equivalent to Ampere’s law. Let’s try to Rindicate
how this comes about. Consider some arbitrary loop for which we want to evaluate B ~ · d~l.
Suppose we “fill in” this loop with a huge number of infinitesimally small loops, all with currents
circulating in the same direction as the original loop (Fig. 10). Then note that adjacent edges
will cancel, since the currents run in opposite directions. Thus we can do our integral by first
computing the integral for all the little tiny loops and then adding the results together. So
3
This is a bit more advanced, and can be regarded as an optional digression.

10
any loop can be built out of tiny loops. The reason this is useful is that it shows that if we
can establish Ampere’s law for a tiny loop, then the result immediately follows for an arbitrary
loop.
So let’s do the line integral for a tiny loop. For simplicity, consider a rectangular loop in
the x-y plane with one corner at the origin (we can always do this by choosing coordinates
appropriately). The integral is a sum of four contributions
Z Z Lx Z Ly Z 0 Z 0
~ ~l =
B·d Bx (x, y = 0)dx+ By (x = Lx , y)dy+ Bx (x, y = Ly )dx+ By (x = 0, y)dy
loop 0 0 Lx Ly
(35)
Because Lx and Ly are very small, we can approximate this by
Z
~ · d~l ≈ [Bx (0, 0) − Bx (0, Ly )] Lx + [By (Lx , 0) − By (0, 0)] Ly
B (36)
loop

For the same reason we can use the following approximations


∂Bx
Bx (0, Ly ) ≈ Bx (0, 0) + (0, 0)Ly , (37)
∂y
∂By
By (Lx , 0) ≈ By (0, 0) + (0, 0)Lx (38)
∂x
Plugging this in gives " #
~ · d~l ≈ ∂By − ∂Bx Lx Ly
Z
B (39)
loop ∂x ∂y
The final step is a bit involved, so I will omit it. The idea is to evaluate the quantity in the
square brackets4 using the field given by the Biot-Savart law. You get zero, unless the loop
happens to enclose some current. If it does enclose the current then you recover Ampere’s law.
This completes our “derivation”.

2.6 Use of Ampere’s law


Ampere’s law is a very powerful method for computing the magnetic fields in problems with a
high degree of symmetry. In particular it is much simpler than using the Biot-Savart law.

Example: infinite straight wire


Let’s turn things around and ask how to rederive the field in this example from Ampere’s
law. Let’s first draw a circular “Amperian loop” of radius r around the wire. By symmetry we
know that the magnetic field is constant around the circle, so we can trivially do the integral
R
on the left hand side of (34) to get B ~ · d~l = 2πrB. Setting this equal to µ0 I immediately
reproduces (11).
Good. But this isn’t quite the whole story. What the above gives us is just the component
~
of B tangent to the loop. Any normal component wouldn’t show up in the integral, and so
can’t be determined from the above analysis. So could there be a radial component, or a
component parallel to the wire? By examining the Biot-Savart law we can conclude that no
such components are present. The law immediately shows that the field is orthogonal to the
current. To rule out a radial component, imagine reversing the direction of the current. The
4 ~
The quantity in the brackets is one component of the “curl” of B.

11
Biot-Savart law says that the field flips direction. But reversing the direction of the current
is physically equivalent to turning the wire upside down, and it’s clear that this can’t flip
the radial direction of the field. The only way out of this paradox is if the radial component
vanishes, which is what we wanted to show.
This is a good lesson: Ampere’s law is really useful after we have used symmetry, physical
arguments, or the Biot-Savart law to determine the basic form of the field. Ampere’s law then
gives us the detailed result.
Now suppose we consider two parallel wires, carrying currents in opposite directions, and
separated by some distance. This is a case in which Ampere’s law is not so useful for determining
the field because we no longer have the rotational symmetry. An example of what not to do
is the following. Suppose we draw a big Amperian loop which encloses both wires. The total
current enclosed is zero, since the two wires cancel each other. So we conclude that the line
integral in Ampere’s law is zero. Should we conclude that the magnetic field thus vanishes
along the loop? No! All we can conclude is that there could be a nonzero field, but its sign
alternates along the loop such that the integral vanishes. In fact, it is easy to find the field for
this example using the principle of superposition. We know the field for a single wire, so for two
wires we just add the results (with a relative minus sign for the opposite direction currents).

Example: Cylindrical wire


Now consider another modification of this problem. Suppose the wire is a hollow cylinder
of radius a, and that the current flows uniformly on the surface of the cylinder. We now ask
for the magnetic field inside and outside the cylinder.
First consider the field inside the cylinder. We consider a circular Amperian loop of radius
r < a. The line integral gives B · d~l = 2πrB. By the same logic as above we know that the
~
R

field is tangent to this loop, with no orthogonal component, and that it takes a constant value
along the loop. But now note that Iencl = 0, since we are inside the cylinder. Ampere’s law
thus implies that B = 0 inside the cylinder.
Outside the cylinder we do enclose all the current. Thus the situation is exactly as before,
µ0 I
and we again have B = 2π r
. Altogether we have that

0, r<a

B(r) = µ0 I (40)
2π r
, r>a

Example: Coaxial cable


Let’s modify the problem again. Suppose the wire is “coaxial”: it has a very thin inner wire
surrounded by a cylinder of radius a. Current I flows up along the inner wire, and the same
current I returns down the cylinder. What is the magnetic field everywhere?
The problem is simple. When we draw an Amperian loop we only have to take into account
µ0 I
the current enclosed. So for r < a we enclose current I, and hence B = 2π r
. For r > a the
total enclosed current is zero due to the cancellation of the two parts, and so B = 0.
The fact that B = 0 outside the coaxial cable is useful for use in delicate electronic instru-
ments, since there are no stray magnetic fields which can interfere with the operation of the
rest of the instrument.

Example: Uniform surface current


Consider an infinite two-dimenesional metal sheet in the x − y plane at z = 0. Let a
uniform current flow in the x direction. The current has a uniform density σ in the sense that

12
Figure 11:

in a transverse distance l there is a current σl. What is the magnetic field resulting from this
current? (Fig. 11)
To solve the problem using Ampere’s law we first need to determine the direction of the
field. Looking at the Biot-Savart law we know that the field is perpendicular to the current, so
the field has no x component. Can it have a z component? No. To see this, suppose that it
had a component pointing the +z direction. Now, consider what would happen if we reversed
the direction of the current. Looking at the Biot-Savart law we see that the field must then
flip direction as well. But if you think about it, this doesn’t make any sense. If we just rotate
the metal sheet by an angle π in the x − y plane then we reverse the current, but this clearly
doesn’t change the z component of the field. Hence the z component must vanish5
We have concluded that there is only a y component, and now we just have to determine
its value. The result will be of the form By (z). Furthermore, the right hand rule applied to the
Biot-Savart implies shows that By (−z) = −By (z).
Let’s draw a rectangular Amperian loop as illustrated. Let the loop extend vertically from
z = −a to z = a, and have a width l. The line integral gives
Z
~ · d~l = −2By (a)l
B (41)

and Iencl = σl. Hence we conclude that By (z) = − σ2 , which is to say


σ
ŷ, z<0

~ =
B 2 (42)
− σ2 ŷ, z>0
Note in particular the field is constant, except for the jump at z = 0. This is analogous to
the electric field in the case of a uniform surface charge, although in that case the field points
purely in the z direction.

2.7 Field of a solenoid


This is an especially important example. Let’s ask the following question. How can we produce
a constant magnetic field out of some arrangement of currents? More precisely, given a finite
supply of wire we try to get a constant field over some finite volume. How do we arrange this?
It’s useful to recall the analogous problem for an electric field. The simplest way to get a
constant electric field is via a capacitor. A capacitor is just two parallel metal plates separated
5
One can also see from the Biot-Savart law that the contributions from y and −y lead to a cancellation in
the z component.

13
Figure 12:

by some distance. We put some charge Q on one plate, and charge −Q on the other plate.
Then, there is a constant field between the plates, and a vanishing field outside. This follows
from Gauss’ law.
In the last example we considered a surface current, which seems analogous to the metal
plate of a capacitor. But there’s a big difference: we can obviously use a finite size plate in
the capacitor, but we can’t have a finite sheet of current, since there would be no where for
the current to go once the sheet ends. The closest analog is to bend the current sheet into a
cylinder, with the current flowing around the cylinder. We now claim that this gives rise to a
constant magnetic field inside the cylinder, pointing parallel to the cylinder axis, and vanishing
field outside the cylinder. A real solenoid is a good approximation to this: we take a long wire
and wind into a tightly packed helix.
Let’s first try to motivate the above claim as follows. First recall the field of a single current
loop (Fig. 4). Now suppose we consider two parallel loops, but displaced along some distance
parallel to the axis. To get the field we just add up the contributions from the two loops. Now
a little examination shows that close to the axis the two contributions reinforce each other (Fig.
12) For larger radii it is a bit more subtle. There is some cancellation between the upper and
lower parts of the loops. The upshot is that the field is weaker outside the radius of the loops
than inside.
Now add more loops, and place them close together. We can further link them up to form
a solenoid, as in (Fig. 13a)
Packing in more and more turns we eventually get to (Fig. 13b) which is an excellent
approximation to our cylinder example noted above. We now want to argue that the magnetic
field is constant inside the solenoid, and parallel to the axis, and zero outside. We will use
Ampere’s law and make reference to (Fig. 14) We take the solenoid to have n turns per unit
length, and to have a current I. Also, we’ll take the solenoid to have infinite length.
We first ask if the magnetic field could have a radial component. No it cannot. For suppose
it pointed out radially. If we instead reversed the current in the solenoid the Biot-Savart
law implies that the field would switch direction, now pointing in radially. But switching
the direction of the current is physically equivalent to turning the solenoid over to reverse its
directions. But this operation clearly can’t flip the radial direction of the field.
Next, we ask if the field could have a component circling around the axis. To see that it can’t
draw a circular Amperian loop centered on the solenoid axis. This loop doesn’t enclose any
current, and so Ampere’s law say that the line integral of the field around this loop vanishes.
But by symmetry the field has a constant value around the loop. Hence the component of the

14
Figure 13:

Figure 14:

15
field along the loop must vanish, as claimed.
We conclude that the field must be parallel to the axis of the cylinder. We now ask for
its precise value. Draw an Amperian loop as illustrated in the figure. The line integral gets
contributions from the upper and lower segments (outside and inside the solenoid), B · d~l =
~
R

h[Bin − Bout ]. On the other hand, the enclosed current is Iencl = Inh. Ampere’s law then tells
us
Bin − Bout = µ0 nI (43)
Note that we haven’t stated the precise locations of the inner and outer segments. Suppose
we shifted one of them up or down by a small amount. The enclosed current stays the same.
Ampere’s law then says that the magnetic field must remain the same. So we conclude that the
field is constant both inside and outside the cylinder. The only place where it is not constant
is at the solenoid itself, where the enclosed current jumps.
Finally, on physical grounds it is clear that the field should go to zero if we go arbitrarily
far away. But since the field is constant outside the solenoid, this implies that it must vanish
everywhere outside the solenoid. (43) then gives us the field inside as

Bin = µ0 nI (44)

So we conclude that a solenoid gives rise to a localized constant magnetic field. This has
many applications. It is of interest to try to generate as strong a field as possible. For reference
recall that the field of the earth is about 5 × 10−5 T . To try to generate a large field we try
to pass as large a current as we can through the solenoid. The trouble is that the wires heat
up due to the resistance. To get around this we can use superconducting wires which have
no resistance provided they are kept sufficiently cold. This gives a so-called superconducting
magnet. These are used in particle accelerators and in high speed Maglev trains. The strongest
field that has been achieved in one of these solenoids is somewhere around 20T or so.
For further comparison we can ask about the strongest magnetic fields observed in nature.
Here we look to astrophysics. Certain stars (neutron stars) can have an incredible density of
charged particles. If these particles start moving around in a coherent fashion an enormous
current will be developed, and this will result in an enormous magnetic field. The way this
motion actually gets set up is complicated, but the result is a field of more than 108 T , far
greater than the strongest laboratory field. Stars with these huge fields are called “magnetars”.
When visiting these, it is best to first remove one’s credit cards from one’s wallet.

2.8 Ferromagnetism
We will be brief here; see the book for a little more info. Let’s give an indication of what’s
going on inside a ferromagnet. Basically, you can imagine the atoms making up the ferromagnet
as being like little current loops. Each atom thus gives rise to a magnetic field, as described
above. In the unmagnetized state the current loops will point in random directions, and so
the atoms will cancel each other out. However, if you apply an external magnetic field to the
material the atoms will all line up with the field and point in the same direction. This gives
rise to a strong magnetic field. In a permanent magnet the atoms will stay aligned even after
the external field is turned off. However, the magnet can get demagnetized either by being hit
(causing the atoms to get jostled into random positions), or by being heated (when heated the
atoms will start moving around and again get randomized).

16

S-ar putea să vă placă și