The Little Textbook

of
Electricity and Magnetism
v.3-2

Peter Meyers

February 27, 2024

Contents

1 Electric Charge, Force, and Field 7

1.1 Coulomb’s Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2 The electric field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2.1 Example: the field of an electric dipole . . . . . . . . . . . . . . . . . 10
1.2.2 Field lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3 Continuous charge
distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3.1 Example: the electric
field of a thin line of
charge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3.2 Example: the electric field on the axis of a uniformly charged disk . . 15

2 Gauss’s Law and Conductors 18

2.1 Conductors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2 Gauss’s Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2.1 Finding E˛ in symmetric cases . . . . . . . . . . . . . . . . . . . . . . 20
2.2.2 Gauss’s Law and conductors . . . . . . . . . . . . . . . . . . . . . . . 24

3 Electric Potential 26
3.1 The electrostatic
potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2 The potential of a point charge . . . . . . . . . . . . . . . . . . . . . . . . . 27

1
Princeton University Physics 104 Spring 2024

3.3 The potential of a system of charges . . . . . . . . . . . . . . . . . . . . . . . 28

3.3.1 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.4 Potential energy of a charge distribution . . . . . . . . . . . . . . . . . . . . 30
3.5 E
˛ from V . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.6 Capacitance and capacitors . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.6.1 The parallel-plate capacitor . . . . . . . . . . . . . . . . . . . . . . . 33
3.6.2 Energy storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.6.3 Dielectrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

4 Current and DC Circuits 36

4.1 Ohm’s Law and resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.2 EMF and circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.3 Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.4 Kirchhoff’s rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.5 RC circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

5 Magnetic Forces and Fields 45

5.1 Magnetic force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.2 Making a magnetic field: Biot-Savart . . . . . . . . . . . . . . . . . . . . . . 46
5.2.1 B
˛ from current in a long straight wire . . . . . . . . . . . . . . . . . 47
5.2.2 B
˛ from a current loop . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.2.3 A solenoid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.3 What about magnets? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.4 Ampère’s Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.4.1 Finding B
˛ using
Ampère’s Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.5 Filling in a few gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

Copyright 2024 Peter D. Meyers 2

Princeton University Physics 104 Spring 2024

6 Electromagnetic Induction 56
6.1 Faraday’s Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.1.1 Lenz’s Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
6.1.2 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
6.2 Changing B
˛ makes E
˛ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
6.3 Inductance and inductors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
6.3.1 Self-inductance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
6.3.2 An LR circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
6.3.3 Energy stored in an inductor . . . . . . . . . . . . . . . . . . . . . . . 65
6.3.4 Mutual inductance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

7 AC Circuits 67
7.1 Oscillating currents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
7.2 Complex numbers
review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
7.3 Solving AC
circuits using
complex impedance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
7.4 The general
procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
7.5 Resonance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
7.6 Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

8 Maxwell’s Equations and Electromagnetic Waves 82

8.1 A problem with
Ampère’s Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
8.2 Maxwell’s Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
8.3 From Maxwell to
electromagnetic waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
8.4 Properties of EM waves in vacuum . . . . . . . . . . . . . . . . . . . . . . . 88

Copyright 2024 Peter D. Meyers 3

Princeton University Physics 104 Spring 2024

8.5 Energy transmitted by EM waves . . . . . . . . . . . . . . . . . . . . . . . . 89

8.6 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

9 Phenomenology of Light as an EM Wave 94

9.1 The electromagnetic spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . 94
9.2 Polarization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
9.3 Reflection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
9.4 Light in transparent media . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
9.5 Refraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

10 Interference and Diffraction of Electromagnetic Waves 100

10.1 Thin-film
interference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
10.2 Diffraction from a
narrow aperture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
10.3 Interference pattern
from two narrow
slits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
10.4 Multiple narrow
slits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
10.4.1 The phasor diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
10.4.2 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
10.5 Diffraction from a
wide slit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
10.6 Diffraction-limited resolution . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
10.7 Appendix: using
phasors to find 2-slit intensity vs. ◊ . . . . . . . . . . . . . . . . . . . . . . . 113

A Core skills 114

Copyright 2024 Peter D. Meyers 4

Introduction

From your study of mechanics, you are cluding plenty of examples, where a normal
now very familiar with the fact that objects textbook would have 40 pages. Dan’s writ-
have mass. Mass (quite oddly) enters physics ing style is relaxed and conversational, with-
in two distinct ways: it determines what an out losing rigor. I will try to maintain these
object will do in the presence of forces via virtues in this book.
Newton’s Second Law, and it provides one of
Because the material is so compressed,
the forces found in nature: gravity. We start
you should not let anything in it slip by
this term by introducing a new property of
you. Not on first reading, but by the end
objects: electric charge. Of course, the ob-
of the week, you should understand every
jects you studied in mechanics, say blocks,
step in every example. When I skip some
had many properties: shape, density, rough-
steps, don’t take that as license to do the
ness, color, and so forth. However, charge
same. You must, with pencil and paper,
turns out to be a more fundamental prop-
work out the missing steps. Initially, there
erty than these, a property of the most ele-
will be reminders1 in the text, but they will
mentary particles composing all the objects
get sparser as the goals of this course become
one studies with mechanics.
familiar to you. With the exception of defi-
The implications of this new property nitions, which must simply be remembered,
will easily keep us occupied for the entire you should think about every point, not un-
semester. Everything you see around you til you’ve memorized it, but until it makes
except gravity (including the fact that you sense to you. If you cannot figure some-
can see) is due to the existence of electric thing out, or if you don’t see the point of
charge. Looking inward, everything you ex- some paragraph, ask about it in class or
perience, think, and feel is mediated by elec- in office hours. That said, not every equa-
tric charges. A semester isn’t enough! tion and point is equally important. To help
you focus, important equations are boxed ,
Author’s note: The idea of having a and there is an Appendix called “Core Skills”
compact, terse but complete, textbook for that calls out a set of talents you should bring
Physics 103 at Princeton University came to the table in exams.
from Prof. Daniel Marlow. He produced a
polished gem, with ten-page chapters, in- 1
Do it!

5
Princeton University Physics 104 Spring 2024

I thank Dan Marlow and Gabriele Monte-

falcone for careful reading and helpful com-
ments.

Copyright 2024 Peter D. Meyers 6

Chapter 1: Electric Charge, Force, and Field
Presumably,1 it was the observation of historical; sometimes using k is more conve-
what we now call “static electricity,” from nient, sometimes ‘0 . The SI coulomb turns
pulling combs through hair and rubbing out to be an awkwardly large unit of charge
shoes on carpets, the former causing the hair – a quick look at Coulomb’s Law shows that
to stand up and the latter leading to enter- the force between two 1-C charges 1 m apart
taining sparks, that led to the notion that ob- is 9 billion Newtons.
jects can become electrically “charged,” and Force is a vector, and the unit vector r̂12
thus the notion of charge. The SI unit for in Eq. 1.1 shows that the force on q2 points
electric charge is the coulomb (symbol:C). along the line connecting the charges. You
will note that Coulomb’s Law is almost iden-
tical in structure to Newton’s Law of Gravi-
1.1 Coulomb’s Law tation. This is useful: a lot of the hard-won
mathematical lore carries over. However,
there are two minor-seeming differences that
change everything. Newton’s gravitation has
F12 a minus sign, so masses, which are always
82 positive always attract. But Coulomb’s Law
ha has no minus sign, so same-sign charges
repel.2 More crucially, electric charges can
if be positive or negative. This means that
an object made of charges (as all objects
Figure 1.1: The force between two electric are) can be electrically neutral, with zero net
charges at rest. charge.
The signs of the charges are handled
Coulomb’s Law states that if two objects
seamlessly by Eq. 1.1: we see that if q1 and
with charges q1 and q2 are a distance r12
q2 have opposite signs, F˛ 12 points opposite to
apart, the electric force of q1 on q2 is
r̂12 , and the force on q2 points toward q1 ∆
opposite charges attract.
˛ 12 = k q1 q2 r̂12 © 1 q1 q2 r̂12 ,
F (1.1)
2
r12 4fi‘0 r12
2 A few notes: Coulomb’s Law is a force
law, and the resulting forces add like any
with k = 1/4fi‘0 = 8.99 ◊ 109 N m2 /C2 . The others. Hence, you should, with a bit of
two ways of writing the constant out front are thought, be able to understand why the force
on the right-hand charge in Fig. 1.2 points as
1
The use of “presumably” will indicate made-up
2
history, not to be taken seriously. Hence the standing-up hair.

7
Princeton University Physics 104 Spring 2024

nomena we will study this term. These

q q
experiments have shown that, besides
Coulomb’s Law, electric charges have two im-
portant, universal properties:
• Electric charge is conserved: the to-
tal amount of charge (positive and neg-
of E 8 of E q
ative, added algebraically) in an iso-
lated system never changes. If the to-
Figure 1.2: Electrical forces add like vectors.
tal amount of charge in a system has
changed, charge has entered or left. This
shown.3 is as fundamental as conservation of en-
In a case like Fig. 1.2, we are assuming q
ergy. g
that the charges are at rest, perhaps held • Electric charge is quantized: all ro
in place by other forces. It turns out that Tcharge found in nature is an integer piFoz
Coulomb’s Law is still valid if the charges multiple of a fundamental charge. Un-
are moving, but in that case there may be til the 1970’s, the fundamental charge
other forces due to the charges, which we will was thought to80
rife rife
be the absolute value of fo
study in due time. For now, we can consider the (negative)
a charge on an electron,roa
the charges to be at rest and our realm is quantity called e and measured to be
called electrostatics. Note also that we are about 1.6 ◊ 10≠19 C.4 This means that
treating the charges as points. a hydrogen atom, with one proton of
charge +e and one electron of charge ≠e
In our lecture demonstrations, we will
is exactly neutral.
take advantage of a marvelous gift from na-
ture: there exist readily-available materials
in which charges are free to move, and others
in which charges remain fixed in place. These 1.2 The electric field
are called conductors and insulators, respec-
tively. Examples of conductors are metals Coulomb’s Law allows us to do all of electro-
and liquid solutions containing ions. Ex- statics using the methods of PHY 103: New-
amples of insulators are glass and plastics. ton’s Laws and the conservation of energy,
This allows us to manipulate charges, moving momentum, and angular momentum. How-
them around and then holding them where ever, we now take what appears to be a de-
we want them. tour and look at the same situation shown
The ability to manipulate charges led to 4
1.602 176 634 ◊ 10≠19 C, exactly now, by defini-
the experiments that revealed all the phe- tion! In the 1970’s, quarks, constituents of protons
and neutrons were discovered. Quark charges come
3
Do it! in units of e/3.

Copyright 2024 Peter D. Meyers 8

 q q DASHED
E vectors
F F surer
q BACKWARDS

rife 80 rife fo
ri ri
Princeton University Physics 104 Spring 2024

q q and we can then write the force on q0 placed

Fo
at any position ˛r0 in the presence of a fixed
F I
ED Fo set of charges qi at fixed positions ˛ri as
80 80
it E82 F
˛ 0 = q0 E(˛
˛ r0 ), (1.5)

with E(˛
˛ r0 ) defined in Eq. 1.4.
Figure 1.3: Electrical forces (still) add like vec-
tors. So, what have we accomplished? It seems
that we have taken a one step process to find
in Fig. 1.2 in a different way. We start by the force on q0 (“Use Coulomb’s Law”) and
formalizing the situation into Fig. 1.3. We turned it into a two-step process (“Find the
have charges q0 ...q2 at positions ˛r0 ...˛r2 , and electric field due to other charges evaluated
we are interested in the force on q0 . Using at q0 , then multiply by q0 ”). We have also
Coulomb’s Law, we can write taken a concrete picture (“Like charges repel,
opposite charges attract”) and turned it into
˛ 0 = k q0 q1 r̂10 + k q0 q2 r̂20 ,
F (1.2) an abstract one (“Force is charge times E”).˛
2 2
r10 r20 Finally, what is the electric field E?
˛

where, you can see graphically, ˛ri0 = ˛r0 ≠ ˛ri . It is common to suggest thinking of E(˛ ˛ r)
We quickly generalize Eq. 1.2 to the force on as the force on a “test charge” q0 placed at ˛r,
a charge q0 with N charges around it, divided by q0 , E(˛
˛ r) = F/q
˛ 0 . This is the con-
crete view, and it has two things in its favor:
N
ÿ q0 qi ÿN
qi It is obviously true, and it shows that E(˛ ˛ r)
F
˛0 = k 2 r̂i0 = q0 ◊ k 2 r̂i0 , (1.3) is a measurable quantity. However, E&M is
ri0 i=1 ri0
an intricate and subtle subject, and it is the
i=1

where, in the second step, we have just abstract path that is more productive.
pulled the common factor q0 out of the We consider E(˛ ˛ r) to be a property of
sum. Though it doesn’t seem terribly earth- empty space determined by (mostly nearby)
shaking, we note that, for a fixed set of electric charges. (“Charges make electric
charges qi at fixed positions ˛ri , the sum de- field.”) An electric field gives a force on an-
pends on the location of q0 (through ri0 ), but other electric charge placed in it. (“Charges
not the value of q0 , that is, it is a function of feel an electric field.”) The notion that empty
˛r0 . We give this function a name: the electric space can have properties is an abstract no-
field at position ˛r0 due to charges qi located tion indeed. A through line of this course
at positions ˛ri is for electric (and later, magnetic) fields to
become more and more real to you.5
N
ÿ qi
E(˛
˛ r0 ) © k r̂ ,
2 i0
(1.4) 5
The abstract can be very real – love and justice
i=1 ri0 come to mind.

Copyright 2024 Peter D. Meyers 9

Princeton University Physics 104 Spring 2024

So, with that in mind, we can discuss

properties of the electric field, which, though
abstract, we have a formula for: Eq. 1.4,
which we can rewrite as
+
N
ÿ ÿN
qi
E(˛
˛ r0 ) © k 2 r̂i0 = E
˛ i (˛r0 ), (1.6)
i=1 r i0 i=1

where the last equality recognizes the ith

term of the sum as the electric field of the
individual charge qi evaluated at the position
˛r0 . We see that electric fields add like forces
do. We refer to this as superposition – we can
find E(˛
˛ r) by finding the electric fields due to
the individual charges and superposing (i.e.,
adding) them.
The individual terms in the sum in Eq. 1.6
show the form of the electric field of a point
charge. For a charge q at the origin, the elec-
tric field at a position ˛r is

˛ r) = k q r̂. Figure 1.4: Electric field of positive charge (top)

E(˛ (1.7)
r2 and negative charge (bottom). The arrow gives
the magnitude and direction of E ˛ at the position
(This is easy to remember, it is just of the arrow’s tail.
Coulomb’s Law divided by the “other”
charge.) The SI unit of electric field has no
specific name – one way to write it is N/C. configuration is called an electric dipole.6 We
We see in Eq. 1.7 that E ˛ points radially away want to find the electric field at a point on
from a positive charge, that is, parallel to r̂ the +x axis (with x > a to keep things sim-
and toward a negative charge (antiparallel to ple) and along the +y axis. Then we want
r̂), and falls off as 1/r2 , as shown in Fig. 1.4. expressions for both fields for x ∫ a and
y ∫ a, respectively.

1.2.1 Example: the field of an 6

Because the term is used in many problems, we
define the electric dipole moment as a vector with
electric dipole magnitude equal to the magnitude of each charge
times their separation and direction pointing from
In Fig. 1.5, two point charges +q and ≠q are the negative to the positive charge. The electric
a distance 2a apart along the x axis. This dipole moment in Fig. 1.5 is then 2qax̂.

Copyright 2024 Peter D. Meyers 10

 a a
−" +" x

Princeton University Physics 104 Spring 2024

a) b) &" y
q
&

&# $! + % !
a a
x
x–a
−" +"
a a &"
Figure 1.5: An electric dipole. −" +" x
&# &

Our first tool

E+ is superposition. At each x+a
point b)of interest, say the locations of the
y
tick-marks in qthe figure, we will find the elec- Figure 1.6: Again, with some suggestions.
tric fields ofE each charge separately and add
them. First, we find the directions of the where the expression for x ∫ a comes from
fields. Without looking $! + at
% ! Fig. 1.6, draw the letting a æ 0 in the denominator, but not
field vectors Eonto
– Fig. 1.5 (or, better, onto in the numerator.8 Even though the total
your own sketch). Youx–a should draw two ar- charge is zero, E
˛ is not zero; the partial can-
rows, representing a a the fields from the E+
two cellation gives a field that falls off faster than
−" +"
charges, at each of the two tick-marks.
E – E x
Let’s 1/r2 .
start with the point on the x axis. We see Now try the same approach on the point
x+a
that both contributions to E(x, ˛ 0) have only along the y3 axis. We note that here the
x components, and that, while the two vec- point is equidistant from the two charges; it
tors point in opposite directions, they are not is the fact that the two fields don’t have op-
equal in magnitude, as ≠q is farther away. posite directions that keeps the fields from
Our next tool is E ˛ for a point charge, Eq. 1.7.
cancelling. Try working this through. I got
Try applying this yourself, using Fig. 1.6. I
get
≠2kqa ≠2kqa
≠q E(0,
˛ y) = î ≠æ î. (1.10)
E(x,
˛ 0) = E
˛ + +E
˛≠ = k q
î+k î. (a2 +y )
2 3/2 y∫a y3
(x ≠ a)2 (x + a)2
(1.8)
Some algebra7 gives Doing this at many points gives the field
map shown in Fig. 1.7.
4kqax 4kqa
E(x,
˛ 0) = î ≠æ 3 î, (1.9)
(x ≠ a ) x∫a x
2 2 2
8
Can you explain why we can and must do it this
7
Do it! way?

Copyright 2024 Peter D. Meyers 11

Princeton University Physics 104 Spring 2024

Figure 1.8: Electric dipole field lines. The ori-

entation of the dipole is the same as in Fig. 1.7.
The dark black circles around the charges are
closely-spaced arrowheads showing the direction
of the field.

Figure 1.7: Electric dipole field. • The field is stronger where the lines get
closer together.
1.2.2 Field lines
These properties originate in Gauss’s Law,
Drawing electric field vectors is awkward. As the subject of the next chapter. We will not
you can see from the figures in the chap- use field lines for anything quantitative – all
ter so far, the magnitudes of the fields, and you need to know about them is in this little
hence the lengths of the arrows, vary wildly section.
in even the simplest of figures. Another rep-
resentation is often used: electric field lines.
These are curves whose tangent at any point 1.3 Continuous charge
is the direction of E. distributions
˛ The field lines for a
dipole, plotted for distances much greater
than the separation of the charges, are shown
in Fig. 1.8.9 Up to now, we have been dealing with point
charges, e.g., “a charge at the origin.”11 We
Fun facts about electric field lines: will now extend our perspective to charged
• Field lines cannot cross.10 objects of notable extent. We will think of
• The lines can begin and end only on 11
Physicist Ed Purcell pointed out a fine irony:
charges. Generations of students have been told that point
charges were an idealization. We now think that the
9
Produced with electrons and quarks that make up matter are actu-
https://github.com/tomduck/electrostatics . ally geometrical points, something we have measured
10
Can you give a simple reason why? Hint: con- to be true down to about a billionth the size of an
sider E˛ at the crossing point. atom.

Copyright 2024 Peter D. Meyers 12

Princeton University Physics 104 Spring 2024

charge as we do mass – we know objects P , which we will initially take as lying along
are made of atoms, but we still approximate the x axis at (x, 0). We consider each bit
macroscopic objects with a continuous mass of the length of the rod dxÕ to be a point
density. We will consider extended objects charge with dq = ⁄dxÕ , where ⁄ = Q/(2a)
to have a charge density, fl, which may vary is the charge per unit length, also known as
with position.12 So a problem may state, the linear charge density. We will find dE, ˛
“A solid sphere has a uniform charge den- the contribution to E˛ at P from each bit of
sity fl = 0.004 C/m3 .” We may also consider charge, then add these fields by integrating.
a thin wire to have a charge per unit length, We first note that all the contributions to
⁄ (C/m), or a surface to have a thin layer of E
˛ will point along the x axis, so we’ll drop
charge ‡ (C/m2 ). the vector signs. Then
We can find the electric field of an ex-
tended charged object using E ˛ of a point k⁄ dxÕ k⁄ dxÕ
dE = = . (1.11)
charge and superposition, much as we did R2 (x ≠ xÕ )2
for the collection of point charges in Eq. 1.6.
We integrate this over dxÕ from ≠a to a,
Rather than a formal development, let’s hack
through an example. ⁄ ⁄ a
k⁄ dxÕ
E= dE = . (1.12)
≠a (x ≠ xÕ )2
1.3.1 Example: the electric You can quickly see that a substitution
field of a thin line of makes this easy to integrate, and that sub-
stitution turns out to be R © x ≠ xÕ ; dR =
charge
≠dxÕ .13
⁄ x≠a
y k⁄ dR
dxʹ E=≠
Q ! x+a R2
3 4
+++++++++++++++++++ 1 1
−$ xʹ a x %& = k⁄ ≠ ,
x≠a x+a
dq 3 4
" 1 1
E(x,
˛ 0) = k⁄ ≠ î. (1.13)
Figure 1.9: Find the electric field of a charged x≠a x+a
thin rod along its axis.
(Do the integral yourself, checking all the mi-
The situation is shown in Fig. 1.9. We nus signs.)
have a thin rod with length of 2a and a total From Physics 103, you know you aren’t
charge Q uniformly distributed along it. We done when you have an answer. Is it right?
want to find the electric field at the point
13
Remember that x is the location of P . It is in
12
Unfortunately, we use the same symbols for some sense a variable, but for the calculation it is a
charge density and mass density. fixed value.

Copyright 2024 Peter D. Meyers 13

4
Princeton University Physics 104 Spring 2024

If so, it should make sense. Start with units. We could continue by writing R, cos ◊, and
It’s easiest to just compare to something you sin ◊ in terms of x, xÕ , and y, e.g.,
know is right. Here, since E = kq/r2 for a
1 21
point charge, noting that ⁄ has units of C/m R = (x ≠ xÕ )2 + y 2 2
, cos ◊ = (x ≠ xÕ )/R.
does the trick. Next, does the field point
away from the positive charge? Finally, what The resulting x integral is ugly, but doable,
does this look like very far from the rod? We but the y integral needs a “trigonometric
know what the answer should be: E ˛ must substitution.”14 It turns out that any phys-
look like that of a point charge. Mathe- ical problem that would require such a sub-
matically, we will be looking at the limit of stitution is simply better done using a real
Eq. 1.13 when x ∫ a. Taking such a limit angle in the problem as the integration vari-
is a skill that comes with practice. Here, for able from the beginning. We thus write ev-
example, just dropping the a’s doesn’t do it. erything in terms of ◊ and y (because y isdE
Instead we manipulate a bit first (fill in the y
constant in the integral), using !
missing steps):
y y "
2a k 2a⁄ kQ R= and dxÕ = dxʹ2 d◊. (1.16)
E = k⁄ 2 2
≠æ 2 = 2 , (1.14) sin ◊ Q sin ◊ q
x ≠a x∫a x x
+++++++++++++++++++
since ⁄ ◊ 2a = Q. The second of these relations ais not obvi-
−$ xʹ x
ous. You can arrive at it by starting with
tan ◊ = y/(x ≠ xÕ ), solving
dq for xÕ and differ-
y ! %& entiating to get dxÕ /d◊. You can also do it
" trigonometrically by using Fig. 1.11, where
Q
dxʹ we note that, as dxÕ æ 0, the two lines go-
q
+++++++++++++++++++
ing to P both have length R and both have
−$ xʹ a x base angle ◊. From the small triangle at the

dq P
Figure 1.10: Find the electric field all around a
charged thin rod. dq
Now we go to the general case – finding Rdq y
q "
the field all around the thin rod, as shown in
Fig. 1.10. Since we will bePadding the vector q
dE’s,
˛ using components seems wise. From
q get dxʹ
the figure, dwe
Rdq Õ y Figure 1.11: Deriving dxÕ trigonometrically.
Õ
k⁄ dx " k⁄ dx
dEx =q cos ◊; dEy = sin ◊. 14
R 2
q R2 A shabby trick that drove me away from math
(1.15) forever.
dxʹ
Copyright 2024 Peter D. Meyers 14

4
Princeton University Physics 104 Spring 2024

bottom, we see that sin ◊ = Rd◊/dxÕ . From Finally, we can take a different limit: let
the large triangle on the right, we see that the rod be infinitely long, still with ⁄ C/m
sin ◊ = y/R. Combining these gives the sec- on it. In that case, ◊1 æ 0, ◊2 æ fi and
ond relation in Eq. 1.16. E
˛ æ ĵ2k⁄/y, a field that falls off slower than
Using Eq. 1.16, that of a point charge.
⁄
k⁄ dxÕ ⁄
y d◊ ⇠
◆ sin⇠2⇠
◊
Ex = cos ◊ = k⁄ ⇠ cos ◊ 1.3.2 Example: the electric
R2 ⇠sin⇠ ◊ y 2
2

k⁄ ⁄ ◊2 k⁄ field on the axis of a uni-

= cos ◊ d◊ = (sin ◊2 ≠ sin ◊1 ),
y ◊1 y formly charged disk
(1.17)
⁄ Õ ⁄ ◊2
k⁄ dx k⁄ a) b)
Ey = 2
sin ◊ = sin ◊ d◊ q q q
R y ◊1 "# "#′ "#
k⁄
= ≠ (cos ◊2 ≠ cos ◊1 ). (1.18) ! z ! z !
y
a
where ◊1 and ◊2 are the angles to the +x axis dq q
x
dq dqʹ
x

of lines from P to the left and right ends of

the rod, respectively.
Figure 1.12: A disk of radius a with a charge q
Check the units. Check that the field distributed uniformly on its top surface.
points away from the positively charged rod.
Check that, along the x axis, Eqs. 1.17 Figure 1.12a) shows a disk of radius a with
and 1.18 give the same answer we got in a total charge q. We will consider the case
Eq. 1.13.15 where the charge is uniformly distributed on
Let’s check the limit of far away along the the top face of the disk, in a layer of infinites- 6

y axis, that is, x = 0, y æ Œ. We see that imal thickness. We want to find E ˛ at an arbi-
along y, Ex = 0 and cos ◊1,2 = ±a/R æ trary height along the axis of the disk, chosen
±a/y since R yæŒ
≠æ y. So, to be the z axis in the figure. As with the rod,
we will find the electric field by superposi-
k⁄ a a 2k⁄a kQ tion, finding the contribution to E ˛ from each
Ey æ ≠ (≠ ≠ ) = 2 = 2 , (1.19) small bit of charge dq, dE ˛ in the figure, and
y y y y y
adding them. Looking only along the axis al-
since ⁄ ◊ 2a = Q. lows us to use the symmetry of the situation
15
to simplify the calculation. The first appli-
Hint: y and both sines are exactly zero on the
x axis, so dealing with Eq. 1.17 is tricky. Note that, cation of symmetry is shown in Fig. 1.12b),
generally, sin ◊1,2 /y = 1/R1,2 , with R1,2 the distance where we note that for each bit of charge
to the respective end of the rod. dq, there is an equal symmetrically-located

Copyright 2024 Peter D. Meyers 15

Princeton University Physics 104 Spring 2024

charge dq Õ . We see that the horizontal com-

q
ponents of dE ˛ and dE
˛ Õ cancel, hence the field
on the axis must point along the axis, and we "#
need only find Ez which is
⁄
k dq ! q z
Ez = cos ◊. (1.20)
R2
a
We are free to divide q into bits dq any dq r x
way we like, with one absolute rule: the in- dr
tegrand must be constant for any dq. The
easiest way to guarantee this is to make the Figure 1.13: A region over which the integrand
volume containing dq infinitesimal in all di- is constant.
rections. We could thus divide the disk into
little squares with area dx dy and do the re- and Eq. 1.20 becomes
sulting double integral (the charge layer is al-
⁄ a
ready infinitesimal in z). However, the rule r dr
Ez = 2fik‡ cos ◊. (1.21)
does not require that the volume containing 0 R2
dq be small per se, and there is room to be
We note that
clever. Figure 1.13 invites us to consider the
1 21/2
contribution to Ez from various points in the R = r2 + z 2 , cos ◊ = z/R, (1.22)
ring of radius r (not small). We see that, as
long as the width of the ring dr is small, all so
points on the ring have the same R and the ⁄ a
r dr
same ◊. We will now use dq to be the charge Ez = 2fik‡z . (1.23)
0 (r2 + z 2 )3/2
on the ring.
We set up the integral by noting that, In this case, a simple substitution works: let
since the charge is uniform across the disk, u = r2 + z 2 , so du = 2r dr or r dr = du/2.
we can define the surface charge density ‡ © Then
q/A = q/(fia2 ). Then dq = ‡ dA with dA du⁄
2fik‡z 2fik‡z --
-a
the area of the ring. Can you see that, Ez = fik‡z = ≠ Ô = ≠ Ô -
u3/2 u r 2 + z 2 -0
with dr understood to be small, this area is A B
dA = 2fir dr?16 Then dq = ‡dA = 2fi‡r dr, = 2fik‡ 1 ≠ Ô 2
z
. (1.24)
z + a2
16
It is important that you have a way of seeing this
that is convincing to you. Here are the beginnings 2) Subtract the area of a disk of radius r from that
of three ways. In all three we take dr to be very of a disk of radius r + dr. 3) Ask yourself how to
small. 1) Imagine snipping the ring in one place and compute the change of area of a disk as its radius
straightening it out to an (approximate) rectangle. changes by dr.

Copyright 2024 Peter D. Meyers 16

Princeton University Physics 104 Spring 2024

By now you know the drill. Does this make

sense? Check the units.17 Look at some in-
teresting limits. Let’s start with z ∫ a, so
we are far from the disk. As with the rod, we
know to expect it to look like a point charge.
We rewrite Eq. 1.24 as
S A T
3 42 B≠1/2
a
Ez = 2fik‡ U1 ≠ 1 + V
z
C A 3 42 BD
1 a
¥ 2fik‡ 1 ≠ 1 ≠
2 z
k kq
= fia2 ‡ 2 = 2 , (1.25)
z z
where, in the second line, we have used a
Binomial Expansion.18
We can turn our disk into an infinite sheet
of charge by letting a æ Œ. Factoring
Eq. 1.24 differently this time,
Q R
z
Ez = 2fik‡ a1 ≠ Ò b
z2
a 1 + a2
3 4
z
æ 2fik‡ 1 ≠ æ 2fik‡, (1.26)
a∫z a a∫z
where we have sequentially neglected (z/a)2
and (z/a) compared to 1. We see that the
electric field of an infinite sheet of charge is
perpendicular to the sheet and is constant.
The result is usually written using the other
Coulomb’s Law constant: E = ‡/2‘0 .

17
Do it!
18
The B.E. is your Friend! (1 ± x)n ¥ 1 ± nx for
x π 1. It works for positive, negative, and non-
integer n. If you are not familiar with it, look it up
and try it with a calculator.

Copyright 2024 Peter D. Meyers 17

Chapter 2: Gauss’s Law and Conductors
This chapter introduces two distinct top- • In a static situation, nothing is acceler-
ics: a class of materials called conductors, ating.
and yet another way to interpret and apply • Therefore, there is not an electric field
Coulomb’s Law called Gauss’s Law. The two present.
topics are not particularly related to each We conclude that: In a static situation,
other, but we will see that Gauss’s Law pro- the field inside a conductor is zero. This
vides much insight on the behavior of elec- is true even if there are electric fields
tric fields in and near conductors. We are, present outside the conductor. (Go back
for now, still staying in the realm of electro- and look through the argument we just made
statics, in which we consider only charges at with this in mind.)
rest.
We can ask how this comes about. The
answer is that the conductor’s charges have
moved to make a new field. Inside the con-
2.1 Conductors ductor, this new field exactly cancels the
original field. This situation is crudely ren-
Matter is made of atoms, which are, in turn, dered in Fig. 2.1.
made of charged particles. A conductor is a
material in which some of the charges, usu-
ally one or more electrons from each atom,
are free to move through the bulk of the ma-
terial, that is, they are not bound to any par-
ticular atom. What makes this interesting Figure 2.1: Left: a uniform field. Right: a con-
is that such materials exist. Metals are the ductor placed in the field. The charges in the
most common example, but there are others, conductor move until the field inside the con-
such as ions in a liquid. ductor is zero.

We ask a very general question: In equi-

The opposite of a conductor is an insula-
librium, that is, in a static situation, with
tor, in which any charge stays where it is put.
no charges moving, what is the electric field
Nature also provides excellent insulators, like
inside a conductor? Surprisingly, logic alone
glass and many plastics. For now, you will
gives us the answer:
see the properties of conductors and insula-
• There is a force on a charge in an electric tors exploited in our lecture demonstrations.
field. Ultimately, they enable the fabrication of the
• If the charge is free to move, it will ac- electronic circuits that have revolutionized
celerate in an electric field. our society.

18
 Princeton University Physics 104 Spring 2024

2.2 Gauss’s Law Of course, d„ is fated to be integrated,

and we define the flux of E
˛ through S as
Gauss’s Law is another way to look at ⁄ ⁄
Coulomb’s Law. It is very abstract, but, „= d„ = E
˛ · dA.
˛ (2.2)
S S
as you’ll see, its power to make hard prob-
lems easy makes it your Friend.1 We start Gauss’s Law is simply stated: For any
by defining a new quantity called the flux of closed surface S (that is, a surface that com-
an electric field through an infinitesimal bit pletely encloses a volume),
of surface with area dA
j
„= E ˛ = Qin ,
˛ · dA (2.3)
d„ © E
˛ · n̂ dA © E
˛ · dA.
˛ (2.1) S ‘0

Here, n̂ is a unit vector normal to the surface where Qin is the total charge inside (enclosed
and, in the final step, we have defined dA ˛ as by) S. An important detail: a closed surface
a sort of “vector area” whose direction is n̂ allows us to choose a positive direction for
and magnitude is the area dA. This is illus- n̂, and thus dA,
˛ unambiguously. We agree
trated in Fig. 2.2, where we show some dA’s to call the outward direction positive.
as part of a larger surface S. Note that the We can test Gauss’s Law with a simple
dot product means that only the component case: the flux of E
˛ through a spherical sur-
of E
˛ perpendicular to the surface contributes
face of radius R with a point charge q at its
to the flux. There is an easily visualized anal- center:
ogous situation: the flow of water. If you j j 3 4
have a rectangular frame of area A, and you 1 q
„= E
˛ · dA
˛ = r̂ · (r̂ dA)
hold it stationary in smoothly flowing water S S 4fi‘0 R2
j
with velocity v, the “flux of water through 1 q 1 q q
= dA = 4fiR2 = ,
the frame” in m3 /s is ˛v · A.
˛ 4fi‘0 R S
2 4fi‘0 R 2 ‘0
(2.4)
E
where in the second step we have used E a
distance R from a point charge and the fact
that both E˛ and dA ˛ point radially outward
(r̂). We see that E˛ and dA˛ are parallel ev-
erywhere on S, making the dot product easy,
and the rest follows.
Figure 2.2: The ingredients for determining the
flux through two infinitesimal patches dA on a We will not prove Gauss’s Law for the
surface S in a uniform electric field E.
˛ general case, but we can see how it works.
In Fig. 2.3a), we generalize the case we just
1
With a capital F. did to an arbitrary surface S Õ containing the
DA DA
b)
Copyright 2024 Peter D. Meyers 19
s

s
Princeton University Physics 104 Spring 2024

DA DA
2.2.1 Finding E
˛ in symmetric
a) b)

Isis cases
s The notion of using Gauss’s Law to find E ˛
is not immediately promising. In the expres-
Figure 2.3: Visualizing how Gauss’s Law works.
The dashed orange lines show the effective area
sion for flux, E˛ is buried in a dot product
of each dA after taking the dot product E
˛ · dA.
˛ which is in turn buried in an integral over an
arbitrary surface. So, prepare to be delighted
when, with practice, you find yourself able to
do some of these cases in your head. There
are two key features in your favor: “arbitrary
charge. Note that for every dA on the sphere surface” means that you get to choose the
S centered on the charge, we can project surface (which we will refer to as the “Gaus-
to dAÕ on S Õ , where the dashed orange seg- sian surface”), and you will choose the sur-
ment shows the effect of the dot product, face to take advantage of the symmetry
which takes the component of dA ˛ Õ parallel to in these cases.
E. The resulting area is proportional to r2 ,
˛
In practice, there are only three situations
while E Ã 1/r2 . Thus dA and dAÕ give the
where this works, but they are broadly ap-
same contributions to their respective inte-
plicable. We will do examples of all three
grals, and the fluxes through S and S Õ are
here, with diminishing details provided in
equal, and thus both equal to q/‘0 . Fig-
each case so you can practice the logic of this
ure 2.3b), sets up the case of q outside S Õ ,
core skill.
which Gauss’s Law says gives „ = 0. You
should provide the step-by-step argument for
this case. Spherical symmetry

The upshot is simple: Gauss’s Law works A spherically-symmetric charge distribution

because projected area and Coulomb’s Law is one in which the charge density fl(˛r) de-
have reciprocal scaling with distance. In pends only on r, the distance from the center,
this sense, Gauss’s Law is a restatement of and not the direction. Examples would be a
Coulomb’s Law. thin spherical shell of charge, a uniform solid
ball of charge, or a case where fl(r) Ã r2 .
The remainder of this chapter will be de- Let’s take on the case of a sphere of radius
voted to applications of Gauss’s Law: using R that has a total charge Q distributed uni-
Gauss to find E ˛ in cases of sufficient symme- formly through its volume. The task at hand:
try and using Gauss to find general (though use Gauss’s Law to find E ˛ “everywhere”,
still electrostatic) results about conductors. which here means for positions r > R and

Copyright 2024 Peter D. Meyers 20

Princeton University Physics 104 Spring 2024

# # #
#! !"
#"
S
dq2
% %
dq1 %

R %′ R !"′ %′ R
Q Q Q
#′ #′

a) b) c)
Figure 2.4: Applying Gauss’s Law to cases of spherical symmetry.

r Æ R. metry.”
Let’s start with r > R. We do not start We can take the symmetry
larger font argument fur-
with any equation; we first take stock of what ther. As shown in Fig. 2.4b), the magnitude
we know about E(˛ ˛ r) just from the situation of E
˛ will be the same for any two points the
as given. If we take Q to be positive, we ex- same distance from the center.
pect that E ˛ will point generally away from We now choose a Gaussian surface to take
it, as shown in Fig. 2.4a).2 Now we consider advantage of what we know about E.˛ Note in
the implications of symmetry. There are two Fig. 2.4c) that at every point on a spherical
popular approaches. The first is practical: if surface S of radius r, E
˛ and dA˛ are parallel
we had to use our direct methods for continu- and the magnitude of E ˛ has the same value,
ous charge distributions (Sec. 1.3), we might which we will call E(r). This means that
note that for every bit of charge dq1 , there j j j
is a symmetrically-located bit dq2 , and thus E
˛ · dA 1
˛ = 2
E(r) dA = E(r) dA
the non-radial components will cancel. We S S S

say “E ˛ must be radial by symmetry.” The 3 4 Qin 5 Q

= 4fir2 E(r) = = , (2.5)
second approach is more philosophical. As- ‘0 ‘0
sume the field really points in a non-radial where we start with the definition of flux
direction, like, say, E
˛ 1 . Imagine rotating the
and the steps are: 1) E
˛ Î dA,
˛ 2) E(r) con-
whole system about the vector ˛r by 180¶ . E ˛1 s
stant over S, 3) dA = A = area of sphere,
has turned into E2 . But the rotation has
˛
4) actual application of Gauss’s Law, and
changed nothing, so E ˛ can’t change. Again,
5) Qin = Q. Remember: only one of these
we conclude that “E ˛ must be radial by sym-
steps3 is automatic – the others are only valid
for this carefully chosen S. We have found
2
Note: I will generally take Q to be positive in that outside a uniform ball of charge, E ˛ is
making figures, but our solutions will work perfectly
3
well for Q positive or negative. Step 4), Gauss’s Law – always true

Copyright 2024 Peter D. Meyers 21

Princeton University Physics 104 Spring 2024

given by perhaps for small r where there is much more

charge outside than inside S. However, this
1 Q ˛ 1 Q
E(r) = ; E(˛r) = r̂ (r > R). is not the case, as we have derived Eq. 2.7,
4fi‘0 r 2 4fi‘0 r2 in which an inward-pointing E ˛ would give a
(2.6)
minus sign. A final note: though the charge
We see that outside a uniform ball of charge
density is discontinuous at r = R (it steps
E
˛ is the same as if all the charge were at
down to zero outside), E(r) is continuous, as
a point at the center. Looking back, we
you can tell by plugging r = R into Eqs. 2.6
see that we never actually insisted that the
and 2.7.
charge be uniform – it just had to be spher-
ically symmetric.4
What about inside the ball, that is, for Cylindrical symmetry
r Æ R ? If you review the steps for r > R
carefully, you will find that all of the argu- In a charge distribution with cylindrical sym-
ments still work just fine: the field is still ra- metry, the charge density depends only on r,
dial (though maybe it points inward – we’ll where in cylindrical coordinates r is the dis-
revisit this), etc. We conclude that a spheri- tance from the z axis, and not the origin.
cal Gaussian surface (now smaller than R) A typical example would be a rod of radius
will still do its magic. In fact, the only R and infinite length with a uniform charge
change to Eq. 2.5 is in step 5: Qin is no longer density inside. It has to be infinite,7 other-
Q. It’s not obvious, but can you see why wise the charge density would depend on z
Qin = Q(r/R)3 for the uniform ball?5 . (The and not only r. Let’s consider a variant of
result for a non-uniform, but still spherically this: a thin, infinitely long wire with a uni-
symmetric, charge density will be different.) form linear charge density ⁄ C/m.
We thus find6 Symmetry again tells us8 the E ˛ must be
1 Qr radial (now in the cylindrical sense), and, as
E(˛
˛ r) = r̂ (r Æ R). (2.7) shown in Fig. 2.5, we see that if we choose a
4fi‘0 R3
cylindrical Gaussian surface S with its axis
The field of a uniform ball of charge is zero along the wire, we get the same useful mir-
at the center and rises linearly. We might acles as found in the spherical case: E˛ Î dA˛
have thought the field would point inward, and E(r) constant everywhere on the cylin-
4
Science history buffs will recognize this as the
der. But we need a closed surface. Mak-
same problem (though with gravity) that Isaac New- ing it infinitely long will cause some trouble:
ton had to invent calculus to solve. And, yes, he got Qin = Œ. So we make our cylinder with
the same answer. an arbitrary length L and the ends under-
5
A slow-but-sure way is to find the uniform charge
density fl = Q/( 43 fiR3 ) and multiply it by the volume 7
Or “approximately infinite”, say close to a very
of S, a sphere of radius r. long charged wire, but far from the ends.
6 8
Do it! Make the argument yourself.

Copyright 2024 Peter D. Meyers 22

Princeton University Physics 104 Spring 2024

$(&′) $(&) $(&)

!" !" $(&) Though conceptually intricate, the actual
!" mathematical steps were all easy. We already
!"′ r r derived this case at the end of Sec. 1.3.1.
rʹ r
Take a look and convince yourself that we
r got the same answer (allowing for different
S notation). Compare the effort involved there
!" and here.
$(&)
L
Planar symmetry
Figure 2.5: Applying Gauss’s Law to cases of
cylindrical symmetry. In a case of planar symmetry, nothing de-
pends on coordinates x or y.9 This means
that charge is in planes or slabs of infinite
stood to be part of S. In support of this, we
extent parallel to the x-y plane. A sim-
note that everywhere on the ends E ˛ ‹ dA,˛
ple example is a thin sheet of charge with a
so even though the still-unknown E(r) is not
constant surface charge density ‡ C/m2 , as
constant over the end, E ˛ · dA
˛ = 0, so we’re
shown in Fig. 2.6. Symmetry requires that
home free:
j ⁄ ⁄ ⁄ E
˛ point perpendicular to the plane and de-
1
E · dA =
˛ ˛ E · dA +
˛ ˛ + pend only on z, the distance from the plane.
It must also have the same magnitude at the
S
cyl side end1 end2
⁄
2 3 same distance z below the plane, but point in
= E(r) dA + 0 + 0 = E(r)Acyl side
the opposite direction.10 . We wish to find E˛
cyl side
a height z above the sheet. A choice of Gaus-
4 Qin 6 ⁄L
5
= 2firLE(r) = = , (2.8) sian surface that takes advantage of these
‘0 ‘0 symmetries is the “pillbox” shown in the fig-
with highlights: 1) the integral must be done ure. It has top and bottom of area A and
piecewise, 2) doing the dot products, 3)
s
E(r) any shape (we’ve chosen it to be round, but
is constant over the cylinder side and dA = it doesn’t matter at all). It extends the same
A, 4) the area of the cylinder, 5) Gauss’s distance z above and below the sheet. Aided
Law, and 6) finding Qin . This gives by the figure as needed, find E(z).
You should get
⁄ ˛ r) = ⁄ r̂.
E(r) = ; E(˛ (2.9) ‡ ˛ r) = ‡ ẑ
2fi‘0 r 2fi‘0 r E(z) = ; E(˛ (2.10)
2‘0 2‘0
Note that the L’s have cancelled, as they
must – surely the electric field cannot de- above the sheet and minus this below. We
pend on the length of an imaginary surface 9
This choice of axes is arbitrary.
we introduced to calculate it. 10
Fill in the arguments for each of these assertions.

Copyright 2024 Peter D. Meyers 23

Princeton University Physics 104 Spring 2024

A z
indicate the Gaussian surface you are us-
a) y ing. Understand the steps needed to extri-
S
x cate E˛ from the dot product and integral,
s
that is, how the chosen Gaussian surface en-
ables this. We will assume that you can
carry out these three cases with ease (and
b) not just remember the results). So, in an
A
$(&′) $(&) exam, we may give you a long rod with a
!" uniform charge density with the understand-
s !"′ z
ing that you will easily be able to find E
˛ at
z
any point inside it, far from the ends.
Figure 2.6: Applying Gauss’s Law to cases of
planar symmetry. 2.2.2 Gauss’s Law and conduc-
tors
got this same result in Eq. 1.26 when we let
the radius of a charged disk go to infinity, and Consider a conductor of arbitrary shape, a
noted that the field is constant in magnitude, random piece of metal. It starts out neu-
independent of the distance from the sheet. tral (Qtot = 0), but we add some charge to
it. (You’ll see us do this often in lecture.) By
the logic of Sec. 2.1, the charge quickly moves
A note on what we want from you about until E˛ = 0 everywhere inside the con-
ductor. Where does that charge end up?
Physics is an attempt to make sense of the Gauss’s Law gives us some insight: choose
natural world at a very basic level.11 In any Gaussian surface S completely embed-
Physics 104, we want you to do the same ded in the conductor (see S in Fig. 2.7). At
thing in miniature: these topics and meth- equilibrium, E = 0 everywhere in the con-
ods should, with practice, make sense to you. ductor, so the flux through S is zero therefore
Gauss’s Law itself is tough to make sense of, Qin = 0. We have just proved that any net
but each step in applying Gauss’s Law to the charge on a conductor at equilibrium
cases of this section should, with practice, must be on the surface. In the figure, S Õ
make sense to you. So: get this – know is a Gaussian pillbox, small enough so that
and understand the logical steps to find E ˛ the surface of the conductor is locally flat
from Gauss’s Law using symmetry. Always and the surface charge density is constant on
11
the conductor’s surface inside S Õ . The sides
To a physicist, “basic” means fundamental, not
“simple” or “easy.” Similarly “elementary.” I am not are short enough that the flux through them
embarrassed that I’ve spent my career doing “ele- is negligible. Since E = 0 anywhere in the
mentary particle physics.” conductor, Gauss’s Law gives E‹ A = ‡A/‘0 ,

Copyright 2024 Peter D. Meyers 24

Princeton University Physics 104 Spring 2024

s !#

Sʹ
S
!=0

Figure 2.7: Applying Gauss’s Law to a conduc-

tor with added charge.

so E‹ = ‡/‘0 , where E‹ is the component

of the electric field perpendicular to the sur-
face of the conductor just outside the con-
ductor. Using the methods of the next chap-
ter (Sec. 3.5), we will show that, just outside
the conductor, the component of the elec-
tric field parallel to the conductor’s surface
is zero. Our conclusion: E ˛ just outside a
conductor is normal to the surface and has
magnitude E = ‡/‘0 , where ‡ is the value of
the surface charge density on the conductor
at that location.
A final note: we saw that E was continu-
ous at the edge of a uniform ball of charge,
even though the charge density fl took a dis-
continuous jump there. Here we see that we
will have a discontinuous jump in E (E = 0
inside and ‡/‘0 outside the conductor) if we
have a volume charge density that is singular
(infinite), as ‡, with a non-zero charge in a
volume of zero thickness, is.

Copyright 2024 Peter D. Meyers 25

Chapter 3: Electric Potential
$(!′)
In Physics 103, after studying forces and
Newton’s Laws, we introduced the concept $(!′)
of energy. Kinetic energy was related to $(!′) "ℓ
"ℓ
forces via the Work-Kinetic Energy Theo- !"
rem, and for certain conservative forces, like !′
"ℓ
gravity, we could introduce a quantity called !!
potential energy. Like gravity, the electro- O
static force is conservative, and we will in- Figure 3.1: Calculating the change in potential
troduce electrostatic potential energy. Since energy due to a force F ˛ when an object moves
Coulomb’s Law and Newton’s Law of Univer- from location ˛ri to ˛rf along a given path via
sal Gravitation are identical in form, all the infinitesimal steps d˛¸. At each location ˛r Õ along
math is the same, with the usual proviso that the path, dU = ≠F(˛˛ r Õ ) · d˛¸.

electric charges can be positive or negative.

Here, we have defined the potential differ-
ence1 between the points ˛rf and ˛ri
3.1 The electrostatic ⁄ ˛rf
E(˛
˛ r Õ ) · d˛¸ = (3.3)
potential V ©≠
˛ri
U/q,

For a conservative force F ˛ on an object, we which no longer depends on q, but is rather

can define the change in potential energy of a function of the positions ˛ri and ˛rf and
the object due to F
˛ when the object is moved depends on the electric field in the space
from an initial to a final position along some surrounding them. We note that this only
path as makes sense if U and V are independent
of the path between ˛ri and ˛rf . This is, in
⁄ ˛rf
fact, the definition of a “conservative force,”
U ©≠ F(˛
˛ r Õ ) · d˛¸, (3.1)
˛ri and it is true for gravity and Coulomb’s
Law. From Eq. 3.3, the unit of V must
with the symbols defined in Fig. 3.1. be joules/coulomb. It has its own name, the
We apply this specifically to the electrical volt, with symbol V. Figure 3.2 shows how
force by having the object be a point charge the definition in Eq. 3.3 is interpreted graph-
q moved from ˛ri to ˛rf in an electric field: ically. Note that V is a scalar not a vector.
⁄ ˛rf A⁄ B This can make it easier to work with than E.˛
˛rf
UE = ≠ q E(˛ ˛ r Õ ) · d˛¸ = ≠q E(˛
˛ r Õ ) · d˛¸
˛ri ˛ri 1
Note: we deliberately dropped the word “en-
= q V. (3.2) ergy.”

26
Princeton University Physics 104 Spring 2024

$(!′)
point, and we take U to be zero when the two
$(!′) charges are infinitely far apart.2 We will find
$(!′) "ℓ
"ℓ the potential energy by finding the potential
!" due to Q at r, then U (r) = qV (r). We find
!′ the potential a distance r from a point charge
"ℓ
!! Q by applying the definition Eq. 3.4. Before
O we start mucking about with integrals, we
Figure 3.2: Graphical view of the computation pause to establish some important intuition.
of potential difference along a particular path.
We will consider both Q and q to be posi-
tive charges.3 We ask: when we bring q closer
As with potential energy, if we choose a to Q, does its potential energy increase or
reference point and agree to use it as ˛ri in decrease? You need a way to see the an-
all calculations, the potential difference be- swer that speaks to you, so stop here and
comes a function of the final position ˛rf only. think about it. Here are some leading ques-
We call this the potential at point ˛rf , which tions: Do the two charges attract, or repel?
we can then refer to as simply ˛r, that is, If you release q, will it accelerate toward Q,
⁄ ˛r or away? After release, what would happen
V (˛r) © ≠ E(˛
˛ r Õ ) · d˛¸, (3.4) to its kinetic energy? What, then, must be
˛rref happening to its potential energy? Or, you
may prefer another approach: do you have
where V = 0 at the agreed-upon ˛rref .
to do positive or negative work to bring q
The upshot: Instead of dealing with elec- closer?
trical potential energy of a charge q at a po-
Releasing q would result in moving to
sition ˛r directly, we will generally find the
lower potential energy (like dropping a rock).
potential V (˛r) first, then U = qV (˛r). We
Since the like charges repel, this means that
have thus made the same abstraction used in
U decreases as r increases, hence U increases
defining the electric field from the Coulomb
as r decreases. Since we have chosen U (Œ) =
force by writing F ˛ = q E.
˛
0, U (r) Ø 0 ’r for like charges. Repeat the
argument, and you will find that U (r) Æ 0 ’r
for opposite charges.
3.2 The potential of a
The same is true for potential. Poten-
point charge tial increases as you get closer to a posi-
2
We have a point charge Q at the origin. We This is the same choice made in dealing with po-
tential energy for Newton’s Law of Gravitation last
ask: what is the potential energy of another semester.
charge, q, a distance r from Q? For this to 3
Our equations will still work for any combination
have meaning, we must agree on a reference of + and ≠.

Copyright 2024 Peter D. Meyers 27

Princeton University Physics 104 Spring 2024

tive charge and decreases as you get closer tem of point charges. However, E, ˛ which we
to a negative charge.4 The reason for our get from superposition, gets messy quickly.
intuition-building pause: it isn’t so easy to Instead we can superpose the potential di-
get the signs right in these calculations. We rectly. If we have a set of point charges Qi ,
will go into it knowing the correct signs. then at each point ˛r Õ along the integration
˛ r Õ ) = qi E
path, E(˛ ˛ i (˛r Õ ), with E
˛ i the field
Back to the potential a distance r from a
point charge Q. Because the potential does due to Qi . Then we can mess with the order
not depend on the path, we may choose one of the sum and integral:
for convenience, and we choose to bring q in ⁄ ˛r ÿ A
ÿ ⁄ ˛r
B
radially from Œ to r. Then E˛ is always anti- V (˛r) © ≠ E
˛ i (˛r ) · d˛¸ = ≠
Õ
E
˛ i (˛r ) · d˛¸
Õ

parallel to d¸. Skipping some of the steps,

˛ Œ i i Œ

⁄ ˛r ÿ ÿ kQi
= Vi (r) = , (3.6)
V (˛r) © ≠ E(˛
˛ r Õ ) · d˛¸
i i ri
˛rref
⁄ r -r
drÕ kQ -- kQ with ri the distance from Qi to the location
= ≠kQ 2
= - = .
Œ r Õ r -Œ
Õ r ˛r.
(3.5)
The same approach applies to a continu-
This is indeed positive (negative) for Q pos- ous distribution of charge, fl(˛r Õ ). Here Qi æ
itive (negative). Try setting up and doing fl(˛r Õ )dV Õ , with dV Õ an infinitesimal volume at
the integral yourself – to get started, what is location ˛r Õ , as shown in Fig. 3.3.
E(˛
˛ r Õ ) · d˛¸ for our chosen radial path? There
are some squirrelly minus signs, but we don’t
agonize over them – just make sure your re- $(!# )
sult has the correct sign at the end. #
!
Finally, the potential energy of a point !"′
charge q a distance r from a point charge
Q: U (r) = qV (r) = kqQ/r.
!′
3.3 The potential of a Figure 3.3: For a continuous charge distribution,
system of charges the superposition sum becomes an integral.

The potential at a location ˛r in or around

The definition Eq. 3.4 also applies to finding a distribution of charge is then
the potential at a location ˛r due to a sys-
⁄
4
To test your understanding, try making the ar- k dq ⁄ kfl(˛r Õ ) Õ
guments yourself.
V (˛r) = = dV , (3.7)
R R

Copyright 2024 Peter D. Meyers 28

Princeton University Physics 104 Spring 2024

with R = |˛r ≠ ˛r Õ |, as can be seen graphically A uniform ball of charge

in Fig. 3.3.
We will now find the potential (with respect
The important conclusion from this sec-
to Œ) everywhere in and around a uniform
tion: we have two Methods to find the poten-
ball of charge Q and radius R. From our
tial of a charge distribution. 1) Integrate the
study of Gauss’s Law, we already know the
total electric field, per the definition Eq. 3.4,
answer for r > R: Since the field outside the
or 2) Superpose the potentials of each bit of
ball is identical to that of a point charge Q
charge, per Eq. 3.7. In practice, 1) is good if
at the center, the potential must also be that
you already know an expression for E ˛ or it is
of a point charge, Eq. 3.5. Inside, Method 2
easy to find, and 2) otherwise.5
looks hard, but we can easily find E ˛ inside
the ball using Gauss’s Law – in fact, we did,
resulting in Eq. 2.7. We thus use Method 1
3.3.1 Examples to continue the potential inward from r = R:
⁄ r
1
A uniformly charged disk V (r < R) = V (R) ≠ E(˛
˛ r Õ ) · d˛¸
R
⁄ r -r
We will find the potential (with respect to 2 kQrÕ Õ 3 kQ rÕ 2 --
= V (R) ≠ dr = V (R) ≠ -
Œ) along the axis of the uniformly charged R R3 R 3 2 -R
A 3 4 B
thin disk of Fig. 1.13. We will pretend we 4 kQ kQ 1 2 2
2 5 kQ 3 1 r 2
= + R ≠r = ≠ .
didn’t already have E
˛ along the axis to prac- R 2R3 R 2 2 R
tice Method 2. Referring to the figure, and (3.9)
taking dq = ‡ dA = 2fi‡r dr as before,
Some highlights: 1) This is an expression
⁄
k dq ⁄ a
r dr for the integral from Πto r, where we al-
V (z) = = 2fik‡ Ô
ready know the integral from Πto R since
R 0 r2 + z 2
Ô -a
- we know the potential outside. 2) Using E ˛
= 2fik‡ r2 + z 2 -
1Ô 0
2 from Eq. 2.7. The minus sign here is a coin-
= 2fik‡ a + z 2 ≠ |z| ,
2 (3.8) flip. We know that the potential continues to
increase as we move inward because E ˛ con-
where the integral was done via u = r2 + tinues to point outward, and we’ll check that
z 2 . Check this by seeing that it gives the that is what we get. 4) We have substituted
expected result as z æ Œ.6 for V (R), and we can see that, for r < R, the
second term is indeed positive. Had it been
5
It is generally easier to deal with a scalar (V ) negative, you should simply flip its sign to
than a vector (E).
˛ Note that by using Eq. 3.7, we the correct one, and say why,7 without try-
have committed to having V = 0 at infinity by using ing to fix the math.
dV = k dq/R.
6 7
Hint: use your Friend the Binomial Expansion. Because physics.

Copyright 2024 Peter D. Meyers 29

Princeton University Physics 104 Spring 2024

∞ ∞
The potential of a uniform ball of charge ∞
is thus r23
q2 q2
Y q3
r12
_
] kQ ,
r 3
for r > R q1 q1 q1 r13
V (r) = 1 22 4
kQ 3 1
_
[R 2 ≠ 2
r
, for r Æ R. a) b) c)
R
Figure 3.5: Assembling a distribution of charges.
(3.10) Each added point charge sees the electric field
This is plotted in Fig. 3.4. of those previously added.

at a time from infinity to their assigned final

location. The first one is free, as there is no
E
˛ present prior to its arrival. As indicated
in Fig. 3.5b), q2 is added with q1 present,
so the potential energy is increased by q2 V12 ,
Figure 3.4: The potential of a uniform ball of where V12 is the potential due to q1 at the
charge of radius R with respect to infinity. We final position of q2 . q3 is added with q1 and
have set kQ/R = 1. q2 present, so, using superposition of the pre-
existing potentials, the potential energy is in-
creased by q3 (V13 + V23 ), and so on. Hence

U = q2 V12 + q3 (V13 + V23 ) + · · ·

3.4 Potential energy of a kq1 kq1 kq2
= q2 + q3 + q3 + ··· (3.11)
charge distribution r12 r13 r23
We note in the last expression that every pair
The potential energy of a charge q moved
of different charges (12, 13, 23,. . .) appears in
from infinity to a position ˛r Õ in the presence
exactly one term.8 We find some enlighten-
of other charges is U = qV (˛r Õ ), with V (˛r Õ )
ment by writing this in a compact way:
the potential at ˛r Õ due to those other charges.
As you learned in Physics 103, this is equal 1 ÿ ÿ kqj 1ÿ
to the work you would have to do to put q U= qi = qi Vnot i (˛ri ),
2 i j”=i rij 2 i
there, and it is thus of practical interest.
(3.12)
We can also ask a different, but related,
where the double sum includes every pair of
question: how much work is needed to as-
charges twice, and we thus divide by two. We
semble a given charge distribution? That is,
identify the sum over j ”= i as the potential
what is the potential energy of the whole dis-
at the position of qi due to all the charges
tribution? We imagine assembling a collec-
tion of point charges qi by moving them one 8
Verify that this still holds when you add q4 .

Copyright 2024 Peter D. Meyers 30

Princeton University Physics 104 Spring 2024

except qi , irrespective of the order of assem- it by taking small steps d˛¸ from some start-
bly, a result captured in the final equality. ing point. Consider a step in the x direc-
Note how simple this final result is. In our tion, d˛¸ = dx î. The dot product in Eq. 3.14
usual fashion, we generalize this to continu- tells us that ˆV = ≠E ˛ x ˆx, where we use
ous charge distributions by our usual replace- the partial derivative notation to indicate the
ment of sum over charges with integral over change in V due to a step in x only.11 This
charge, tells us that E
˛ x = ≠ˆV /ˆx, and, doing the
same thing in y and z, we see exactly what
1⁄ 1⁄ kind of derivative of V E ˛ is:
Uæ V dq = V (˛r Õ )fl(˛r Õ ) dV Õ .
2 2 A B
(3.13) ˛ = ≠ ˆV î + ˆV ĵ + ˆV k̂ .
E (3.15)
ˆx ˆy ˆz
As with finding potential, there are again
two Methods9 for finding the electrostatic So, given V as a function of (x, y, z), we can
potential energy of a charge distribution: 1) find E(x,
˛ y, z) everywhere.12
assemble the distribution mathematically, or
2) use Eq. 3.13. And again, often one is We won’t use Eq. 3.15 much in this course,
tractable when the other isn’t. As an ex- but we will use Eq. 3.14 directly to build our
ercise, try finding the potential energy of the conceptual understanding of the relationship
uniform ball of charge of Sec. 3.3.1 using both of E
˛ and V . Consider again a non-uniform
methods (both of which are tractable).10 electric field E(x,
˛ y, z) and a step d˛¸ of fixed
length but unspecified direction. From the
property of the dot product, we see that if
we want to make the biggest change in V
3.5 E
˛ from V with our fixed-length step, we should take it
in the direction of E˛ (or its opposite): The
Since V is some kind of integral of E ˛ electric field always points in the direc-
(Eq. 3.4), E˛ must be some kind of deriva- tion of steepest decrease of V . Similarly,
tive of V . From the integrand in Eq. 3.4, we A surface everywhere perpendicular to
see that E
˛ 13 is a surface of constant potential,
dV = ≠E ˛ · d˛¸. (3.14) 11
Partial derivatives appeared briefly in PHY103
All information about the relation between E
˛ in the Wave Equation. Don’t be concerned if you
haven’t seen partial derivatives in math yet. For a
and V is encoded here – we just need to de- function V (x, y, z), ˆV /ˆx is just the derivative of V
code it. We will consider some non-uniform with respect to x keeping y and z fixed, as we are
electric field E(˛
˛ r) = E(x,
˛ y, z) and explore doing here.
12
If you’ve had some vector calculus, you will rec-
9
There are others as well, see Sec. 3.6.2. ognize this as E˛ = ≠ÒV ˛ , that is, E ˛ is minus the
10
Hint: in both methods you end up with dq being gradient of V .
13
a thin spherical shell of charge in the ball. Take a moment to visualize this.

Copyright 2024 Peter D. Meyers 31

Princeton University Physics 104 Spring 2024

since E˛ · d˛¸ = 0 on that surface. This sur- nent of E

˛ normal to the surface is E‹ = ‡/‘0 .
face is called an equipotential surface or just We promised that we would show that EÎ =
an equipotential. Another way to say this is: 0 later, making E
˛ = ‡/‘0 n̂, with n̂ the nor-
Field lines are perpendicular to equipo- mal to the surface. Later is now.
tentials. This is all summarized in Fig. 3.6.

3.6 Capacitance and ca-

pacitors
i 40 V Consider a conductor of arbitrary shape14 ,
60 V
with a net charge Q. It is an equipotential,
100 V
80 V
and therefore its potential with respect to
infinity is a number V that applies to any
point in the conductor. While it may be dif-
Figure 3.6: The relationship among E ˛ (red ar-
ficult to calculate this number, we can easily
rows), electric field lines (dashed red lines), V
(blue labels), and equipotentials (blue lines). see that V Ã Q. The equilibrium distribu-
tion of the charge added to the conductor
is the one that gives E ˛ = 0 everywhere in-
We can also use the conceptual under- side. Scaling this charge distribution up by
standing of potential to improve our intuitive some constant factor, giving QÕ = f Q still
understanding of conductors. gives E˛ = 0 everywhere inside, and so that
is how a total charge QÕ would distribute.
• Since, in a static situation, E
˛ = 0 every-
Changing Q to QÕ would thus scale up E ˛ by f
where inside a conductor, the potential
everywhere outside the conductor, and that
difference between any two points in the
results in V Õ = f V . So for a conductor of
conductor is zero. (We can choose an
any shape, we can write Q = CV , with C
integration path for Eq. 3.3 to be en-
a constant that depends only on the geome-
tirely within the conductor, so E ˛ = 0
try (shape and size) of the conductor. C is
everywhere on the path.) Hence, in a
called the capacitance of the conductor, and
static situation, a conductor is an
it is defined as C © Q/V . The SI unit of
equipotential.
capacitance is the farad (symbol: F), with
• Since the electric field is always per-
1 F © 1 coulomb/volt.
pendicular to equipotentials, the elec-
tric field just outside a conductor We can generalize this to a pair of con-
is normal to the surface. ductors of arbitrary shape. Though we could
In Sec. 2.2.2 we used Gauss’s Law to show 14
Which somehow always seems to be the Green
that, just outside a conductor, the compo- Potato of Fig. 3.3.

Copyright 2024 Peter D. Meyers 32

 Princeton University Physics 104 Spring 2024

put arbitrary charges on each conductor and plates as infinite sheets. The goal is to find
compute each conductor’s potential with re- the capacitance of this setup.
spect to infinity, the most common applica-
tion is to: Following the three-step scheme above, we
imagine that we’ve moved a charge Q from
1. Start with each conductor neutral. the bottom plate to the top plate. The fig-
2. Move a charge Q from one conductor ure shows all the charge on the inner surfaces
to the other, leaving the first conductor of the plates. This is intuitively correct: the
with ≠Q and the second with +Q. +Q and ≠Q attract, and they are free to
3. Determine the potential difference be- move inside the conductors. Confirm that
tween them, V © V+ ≠ V≠ , which we this configuration indeed gives E ˛ = 0 inside
will simply call V , a positive number by each plate. Given the known charge distri-
definition. bution, we calculate V to find C. To find
In this setup, the pair of conductors is called V , we can first find E.˛ This is most eas-
a capacitor and C © Q/V , with Q the (pos- ily done by using E = ‡/‘0 just outside a
itive) charge on the positive conductor. conductor and the knowledge that the field
from an infinite sheet of charge is constant,
+
Eq. 2.10. (You can get the same result by su-
3.6.1 The parallel-plate capac- perposing the electric fields from two oppo-
sitely charged sheets using Eq. 2.10.) Then,
itor using a path straight up from the bottom
plate to the top plate,
A ++
++
+Q + ++
+ ++
+++++++++++++++++++++++++++++++++ +
⁄ ⁄ d
‡ Qd
d !
−−
−−
−
V = ≠ E·d
˛ ˛¸ = E dz = Ed = d= ,
−Q −− 0 ‘0 ‘0 A
−−
z=0 −−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−− −
−
(3.16)
Figure 3.7: A parallel-plate capacitor. Not to where we have used ‡ = Q/A. So,
scale: we will consider the (realistic) case where
d is much smaller than the length or width of a
plate. ‘0 A
C © Q/V = . (3.17)
d
The simplest and most common capaci-
tor is the parallel-plate capacitor shown in
Fig. 3.7. The two conductors are flat plates, Since capacitance tells you how much charge
each with area A, separated by a distance d. you have to move to get a given V , think
We will be interested in the case when the about it until it makes sense to you that C
separation is much
Ô less than the size of the is bigger for bigger area and for smaller sep-
plates, say d π A. We will then treat the aration.

Copyright 2024 Peter D. Meyers 33

Princeton University Physics 104 Spring 2024

3.6.2 Energy storage plate cap, we can write

3 4
1 1 ‘0 A 1
A capacitor is a repository for separated +
15 U = CV 2 = (Ed)2 = ‘0 E 2 (Ad).
2 2 d 2
and ≠ charge. It took work to separate those (3.20)
charges, and from a physics perspective, a Since Ad is the volume of our cap, the impli-
capacitor stores energy. We can easily cal- cation is that there is an energy density
culate this potential energy using Eq. 3.13,
since all the ≠Q is at V = 0 and all the +Q 1
uE = ‘0 E 2 (3.21)
is at V . 2
in the cap. This turns out to be a general
⁄
1 1 1 1Q 2
result, and gives the very abstract view that,
U= V dq = QV = CV 2 = ,
2 2 2 2C instead of viewing this energy as the poten-
(3.18) tial energy of an array of charges (Eq. 3.18),
where we have shown the result in three use- or the work done to assemble the array
ful forms.16 (Eq. 3.19), the energy somehow resides in the
electric field.
We can use our other Method, assembly
of the charge distribution, for practice. We
consider an arbitrary moment in the transfer 3.6.3 Dielectrics
of charge from the negative plate to the pos-
itive one. At this moment, there is already a Real capacitors are generally not two con-
charge QÕ on the top plate, and the potential ductors separated by vacuum. To boost ca-
of the top plate is V (QÕ ). We find the work pacitance (for example, to have more stored
needed to move an additional dQÕ , which is energy at a given voltage) and for practical
then the increase in potential energy stored reasons (to keep the plates from touching as
in the cap: dU = V (QÕ ) dQÕ and d is made as small as possible), an insulator,
also called a dielectric, is usually present be-
⁄ ⁄
QÕ 1 Q2 tween the plates. To see why simply having
U= V (Q ) dQ =
Õ Õ
dQ =
Õ
, a dielectric present increases C, we must re-
C 2C
(3.19) alize that a dielectric cannot be completely
as we got previously. inert. Even if no charges are free to move
around in the bulk of the material (what we
Though we won’t attempt a general
mean by an insulator), the dielectric is still
derivation, we note that, for our parallel-
made of atoms, and the atoms have charged
15
constituents, which feel a force in an electric
A “cap” to its friends
16
It should be noted that the amount of energy field. In the simplest case, the electrons and
stored is generally small. Try C = 1 µF (a typical nuclei are pulled in opposite directions, “po-
value) and V = 10 V in Eq. 3.18. larizing” the atoms into dipoles. Consider

Copyright 2024 Peter D. Meyers 34

Princeton University Physics 104 Spring 2024

our parallel-plate capacitor, still with charges

±Q on the plates, but now filled with dielec-
tric (see Fig. 3.8). If all the negative charges
are shifted up slightly and all the positive
charges are shifted down slightly by the elec-
tric field, the result is that the bulk of the
dielectric remains neutral, but there is a net
+ charge at the bottom and ≠ at the top.
This creates its own E ˛ Õ opposing the origi-
nal E˛ due to ±Q.
A ++
++ −
+Q + +−
++
++ −
+++++++++++++++++++++++++++++++++ +
− − − − − − − − − − − −
+
d ! !′ + −−
−
−
−Q + −−
−
+ + − −−
z=0 + + + + + + + + + +
−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−− −

Figure 3.8: A capacitor with dielectric.

The result is a smaller net V in the cap

for a given Q, that is, a bigger capacitance.
This system turns out to behave linearly in
many practical cases, and E = E0 /Ÿ, with
E0 the field in the same capacitor with the
same Q, but no dielectric. Ÿ17 , a dimension-
less number > 1 and typically in the range 1-
10, is called the dielectric constant, and it is a
property of the dielectric material. The same
calculations leading to Eq. 3.17 now give
Ÿ‘0 A
C= . (3.22)
d

17
The Greek kappa.

Copyright 2024 Peter D. Meyers 35

Chapter 4: Current and DC Circuits
We now relax the requirements of elec- This can be written as
trostatics and turn to the study of moving
charges. We will focus here on charges in Qpast P in qn(vd t)A
I= = = qnvd A,
t

conductors, specifically wires and other com- t t

(4.1)
ponents that make up electric circuits. We
where n is the number density (number per
have said that a conductor in a static con-
volume) of the charges, and the volume con-
dition is an equipotential, with E ˛ = 0 in-
taining the charges that will pass P in the
side. Here we will disturb that equilibrium
next t is vd tA.
and contrive to have a potential difference1
across the conductor, and E ˛ ”= 0 inside. The We know that in most conductors the
free charges inside will thus move: an electric moving charges are electrons, with q = ≠e =
current. ≠1.6◊10≠19 C.2 Though it is odd to think of
it this way, the bookkeeping works out best
when we treat negative charges mov-
CHARGESWHERE
p
WILLPASSPIN
THENEXTot 1 ing in one direction as a current in the
opposite direction. So, if the charges in
A Iliff o
Fig. 4.1a) are positive, the current is to the
Nd ICitqso right, but if they are negative (e.g., electrons)
8
r
x
lmeaar.GE V0L the current is to the left.
vdot Ten amperes is a hefty current (a typi-
Figure 4.1: Defining the current for charges cal household circuit breaker trips at 15 A).
flowing in a wire. With some easy-to-find properties of copper,
I and the less-easily found fact that each cop-
Iast
We quantify this by specifying the charge per atom contributes one free electron to
per second passing any point along the con- a copper wire, we can find the drift speed
ductor. The SI unit isREGION
THIS C/s, called the ampere of electrons in a 2-mm-diameter (A ¥ 3 ◊
(symbol A). Figure8 4.1 shows
_I Iz 0 the situation in 10≠6 m2 ) copper wire carrying a 10-A cur-
a wire with cross-sectional area A. The free rent. With I, q, and A known, this comes
charges q move with an average velocity vd , down to finding n. We know that each
called the drift velocity. (The figure shows Cu atom contributes one free electron, so
all the charges having vd for simplicity; the we have to find the number density of cop-
reality is quite different.) The current is the per atoms. Looking up the density and
total charge passing the point P in t = 1 s. atomic (molar) mass of copper (9 g/cm3 and
2
Note that we define e as a positive number, a
1
Often called a voltage in this context. common practice.

36
Princeton University Physics 104 Spring 2024

63.5 g/mol, respectively) we get3 notable exception is systems that include the
capacitors we introduced in Sec. 3.6.1. We
9 g/cm3 will see their effect later in this chapter. In
n= (6◊1023 Cu/mol)(1 e/Cu)
63.5 g/mol this chapter, we will also generally be consid-
= 8.5◊1022 e/cm3 = 8.5◊1028 e/m3 . ering steady currents, that is, currents that
(4.2) are not changing over time. Again, an excep-
tion will be systems with capacitors. Later
Then, solving Eq. 4.1 for vd , in the course, we will expand our study to
include systems where the current oscillates.
I
vd = = Though we will not refer to it again in
enA p
CHARGESWHERE
WILLPASSPIN
101 C/s this book, some assigned problems may re-
THENEXTot
fer to a quantity called current density using
(1.6◊10≠19 C)(8.5◊1028 m≠3 )(3.1◊10≠6 m2 )
the
o symbol ˛ J. We have defined current a bit
= 2.4◊10
A
≠4
m/s = 0.24 mm/s. Iliff
(4.3) loosely, as “charge passing a point” per sec-
Nd ICitqso
r
x
8 speed ond. In Fig. 4.1 we see that we really meant
So, in this BIG current,lmeaar.GE
the net of the
V0L the charge going through a cross section of
electrons is about a quarter of a millimeter
vdot the conductor. Current density is charge per
per second. Is that what you expected?
second per area measured at any point, with
vector ˛J carrying the direction of the current.
For current density uniform across the cross
I Iast section of the conductor (a common case, and
sufficient for our purposes), I = JA.
THISREGION
8 _I Iz 0
Figure 4.2: If the current is not the same at 4.1 Ohm’s Law and resis-
every location along the wire, charge must be
building up or dissipating.
tance
OK, we apply a potential difference V
Figure 4.2 illustrates another property of
across an object made from a conducting ma-
current: if the currents at different loca-
terial, and we expect a current I to flow.
tions along a single wire differ, there must
What I will we get for a given V ? Ex-
be charge building up or dissipating some-
periment shows that, for objects made from
where in between. This is just conservation
many conducting materials, I Ã V . Such
of charge. We will generally assume that this
materials are called Ohmic because they
is not happening in the systems we study. A
obey Ohm’s Law:
3
If this is unfamiliar to you, ask about it in office
hours. I = V /R, (4.4)

Copyright 2024 Peter D. Meyers 37

Princeton University Physics 104 Spring 2024

where R, the resistance, is a constant for that We have now arrived at the situation of
particular object. The SI unit for R is the Fig. 4.3a), a simple application of Ohm’s
volt per ampere, which is called the ohm, Law, with the notation changed a bit in
with symbol . Fig. 4.3b). Our immediate goal is to take
R depends on the resistivity, fl, of the ma- the leap from this situation to the analysis
terial4 , and the geometry of the object. For of practical electrical devices which we will
a V placed across the length L of an object call direct current (DC) circuits. There is
of uniform cross-sectional area A, a lot of lore and jargon involved, but all of
it comes from 1) the definition and meaning
of potential and current, 2) Ohm’s Law, and
R = flL/A. (4.5)
3) restricting ourselves to the position- and
time-independent steady currents described
This is easy to remember, because it fol- above.
lows our intuition for water in a hose. For a
given water pressure, we expect a higher flow Consider the case of Fig. 4.3c): two re-
(more current, lower resistance) for a bigger- sistors connected as shown. This end-to-end
diameter hose (more “room”), and a lower configuration is called a series combination.
flow for a longer hose (more “friction”). We We want to “analyze” or “solve” this config-
see from this relation that fl has units of ·m. uration, which generally means to find the
The resistivity of common materials ranges current through and the potential difference
over 22 orders of magnitude5 with copper (called the “voltage drop” when we go in the
(an excellent conductor), carbon, and glass direction of the current) across some or all
(an excellent insulator) having fl ¥ 2◊10≠8 , of the components (in this case, the resis-
fl ¥ 4◊10≠5 , and up to 1014 ·m, respectively. tors). We start by defining a new parameter,
Resistance is a way of manipulating current VP , the potential at the point connecting the
and, as we will see, energy, in electronic de- resistors. Defining V1 and V2 as the voltages
vices. An electrical component called a re- across the two resistors6 and applying Ohm’s
sistor is used to deliberately introduce resis- Law, we have
tance. Equation 4.5 tells us that if we want
V1 © VP ≠ V0 = ≠I1 R1
to make a 1- resistor that is a cylinder 3
mm in diameter and 10 mm long, we need V2 © 0 ≠ VP = ≠I2 R2 . (4.6)
a material with fl = 7 ◊ 10≠4 ·m. We can Take a minute to understand the minus signs.
compare this to the resistance of 10 cm of The current flows from higher to lower poten-
1-mm-diameter copper wire, which, from the tial. This could either be a current of pos-
same equation, is 0.003 .
is the ratio of the width of your hand to the size of
4
The world uses fl for both resistivity and charge our galaxy.
6
density – BEWARE. Note that it is common to use V interchangeably
5
This is quite remarkable. 22 orders of magnitude as a potential or a potential difference – sorry!

Copyright 2024 Peter D. Meyers 38

Princeton University Physics 104 Spring 2024

$! = ∆#/! $# = #" /! $$ $% $& $$

$( $(
! ! !$ !% !' ! !' !$
!%
!!
∆# #" #=0 #" 0 #" 0
#" $% 0
a) b) c) d) e)
Figure 4.3: The zig-zag line is the symbol for a resistor. a) Ohm’s Law. b) Identifying one point as
V = 0. c) Two resistors in series. d) A resistor and two wires in series. e) Two resistors in parallel.

itive charges flowing to the right, or nega- the resistor with a wire at each end. This is
tive charges flowing to the left. In most real three resistances in series (R and both wires,
cases, it is the latter. However, it is per- each seeing the same current), so the resis-
fectly good to pretend it is the former, with tances add and Id = V0 /Req . But, as we saw,
no fudging of signs needed at the end. In the resistance of a typical wire is negligible
this application, the two cases are the same. when added to that of a typical resistor, so,
Continuing to deploy our short list of prin- to a good approximation, Id = Ib = V0 /R
ciples, the restriction to steady currents tells – we can neglect the presence of the wires
us that I1 = I2 , and we will call them both (called “leads”) that connect components to-
Ic . We also know (from the meaning of po- gether.8 We can use Ohm’s Law again and
tential, or from Eq. 4.6) that V1 + V2 = ≠V0 . conclude that the potential difference across
Hence each wire is Vw = Ib Rw = V0 (Rw /R), which
is also negligible.
≠ V0 = V1 + V2 = ≠Ic R1 ≠ Ic R2
∆ V0 = Ic (R1 + R2 ) © Ic Req With wires at our disposal, we can make
the configuration shown in Fig. 4.3e): two re-
∆ Ic = V0 /Req , with Req = R1 + R2 , sistors in parallel. From our analysis of case
(4.7) d), we will look for Req of parallel resistors.
where Req is the equivalent resistance of the Noting that the voltage drop in the wires is
two resistors in series. We have solved this negligible, we see that each resistor has V0
case, but we have also found a useful result: across it9 , hence I1,2 = V0 /R1,2 , and the cur-
resistances in series add.7 rent is bigger through the smaller resistor.
We now take a look at the nodes where three
We immediately use our conclusions from wires meet, indicated by dots. Charge con-
Fig. 4.3c) to analyze Fig. 4.3d). Here we have servation and our requirement that there be
7
It is typical to consider series to mean “end-to-
8
end”, but, looking at our derivation, the crucial point If higher precision is needed, we could replace R
was that components in series see the same current, with Req and then neglect the wires.
9
and “see the same current” is a better definition of So our physicsy definition is components in par-
“series”. allel see the same potential difference

Copyright 2024 Peter D. Meyers 39

Princeton University Physics 104 Spring 2024

no charge buildup require that the total cur- "

rent Ie = I1 +I2 . (We will use this Node Rule ! !
a lot.) Then

+
V0 V0 V0 ℰ
= Ie = I1 + I2 = +
Req R1 R2 Figure 4.4: A simple DC circuit. At the bottom
3 4
1 1 is the symbol for a battery providing an EMF E.
= V0 +
R1 R2 Sometimes the + sign isn’t included; the positive
1 1 1 end has the longer line.
∆ = + (4.8)
Req R1 R2
quantity conveniently measured in volts. A
for resistors in parallel. device that does this via chemical reactions
is a battery.11
Figure 4.4 shows the simplest DC circuit,
4.2 EMF and circuits with DC standing for “direct current,” mean-
ing current that flows in only one direction.
To have a steady current flowing in any of We solve such a circuit by deploying the same
the Fig. 4.3 cases, we need a way to provide principles used in the previous section, ap-
an unchanging potential difference V0 . We plied to the loop as a whole. The current
will generally do this by adding a new device must have the same value, I, everywhere, as
to complete an electric circuit in which the shown. We can choose any point in the cir-
current flows in a closed loop, through the cuit as V = 0 and follow the loop, account-
new device, then through the resistors, and ing for the changes in potential. When we
back to the device again. Noting that inside arrive back at our starting point, we must
the device, the current goes from low poten- be back at V = 0 again. This Loop Rule
tial (zero in the figure) to a higher potential thus tells us that the total change in poten-
(V0 ), we realize that a simple electric field in- tial around the loop, counting EMF’s as if
side the device won’t do this (current moves they were potential differences, is zero. In
in the direction of E,˛ and E ˛ points toward Fig. 4.4, we can start at the lower-right cor-
lower potential.) What we need is a device ner and travel clockwise with the current,
that can increase the potential energy of the giving +E ≠ IR = 0. For the given E of
charges by means other than a static elec- the battery and R of the resistor, this gives
tric field making a potential difference. We us the current I = E/R.
call this a device that provides an EMF10 , a
11
We won’t be concerned with how a battery does
10
This used to stand for “electromotive force,” but, this in PHY 104, but look it up sometime. Not to
since it doesn’t even have units of force, we now just be outdone by the chemists, in the second half of the
say “Ee Em Eff.” course we will see how to make an EMF with physics.

Copyright 2024 Peter D. Meyers 40

Princeton University Physics 104 Spring 2024

A final note on batteries: the EMF of real and the

batteries can sag a bit if it provides a large Node Rule: the sum of currents enter-
current. A simple model of this assigns an ing any node is zero.
internal resistance to the battery, which acts
as a resistor Rint in series with the “ideal” are called Kirchhoff’s Rules. These turn
EMF E. The voltage across the terminals of complicated circuits into algebra problems:
the real battery when it provides a current I each loop and each node is an equation. Be-
is thus E ≠ IRint . Unless stated otherwise, ing able to solve these circuits (find all the
we will pretend that all batteries are ideal, currents and the voltages across the compo-
that is, Rint = 0. nents) is a useful (and required!) skill, which
will come with practice.
10 W N1 4W

4.3 Power 10 V
!$ !(
!% 2W
L1 L2
4V
A resistor R with current I through it has
a potential V = IR across it, so I coulombs
per second pass through the resistor, losing V N2 2W
joules per coulomb of potential energy. Thus Figure 4.5: A multi-loop DC circuit.
I ◊ V J/s are lost to the current. This en-
ergy appears as heat in the resistor (and is We will try this out on the example in
restored to the current by the battery, which Fig. 4.5, using this recipe:
drains in energy.) A J/s is a watt: the bat- 1. If there are obvious parallel combina-
tery is delivering power P = IV to the re- tions of resistors, replace them with
sistor, which dissipates it as heat. Though their Req ’s.
this can be an annoyance, it also gives us our 2. Give names and directions to all cur-
first applications: a toaster, a coffee-maker, rents. These are the “unknowns”. The
baseboard electric heat, and, if you let the re- directions are arbitrary, so just guess. If
sistor get hot enough, an incandescent light you guess wrong, the current will come
bulb. out negative – no problem.
3. Pick a node, and write down its equa-
tion.
4.4 Kirchhoff’s rules 4. Pick a loop and an explicit direction to
traverse it (either way is fine), and write
down its equation. This is a case where,
Together, the if numbers are given, it is better to use
Loop Rule: the sum of potential differ- them, simplifying as you go, rather than
ences around any loop is zero. using symbols like R5 .

Copyright 2024 Peter D. Meyers 41

Princeton University Physics 104 Spring 2024

5. Continue with 3 and 4 until you have were correct, but I2 is the opposite direction
enough equations. Solve. of its red arrow.
6. Check your solutions in the origi-
nal equations.
7. Use the derived currents to find voltages 4.5 RC circuits
across components as needed.
You will get a lot more practice with this, a) b)
in class, in lab, and in problem sets. For now, +&! +&(#) !(#)
I will flesh out the recipe a bit, but leave the " ! " !
L1
algebra to you. In the figure, the red ele- −&! −&(#)
ments are those added by the student (OK,
me) in Steps 3 and 4. Do you see that the
three currents defined by the arrows specify Figure 4.6: An RC circuit. a) With the switch
(new symbol at the top) open so no current can
the currents everywhere? For node N1 and
flow, the capacitor is charged to have +Q0 on
loops L1 (starting at the lower-left corner)
the top plate and ≠Q0 on the bottom plate. b)
and L2 (starting at the lower-right corner), I At a time t after the switch is closed, the charge
get the equations on the plates is ±Q(t).

I1 + I2 + I3 = 0 (4.9)
We now add the capacitor (Sec. 3.6.1) to
+10 ≠ 10I1 + 2I2 = 0 (4.10)
our repertoire. Here, the interesting behav-
+2I3 ≠ 2I2 + 4I3 ≠ 4 = 0. (4.11) ior is time-dependent. We will allow charge
to change on the plates of the capacitor,
Check that every sign in each of the three but nowhere else in the circuit, and we will
equations makes sense to you. A current en- let currents and voltages be time-dependent.
tering (leaving) a node is positive (negative). The simplest case is the RC circuit shown in
The potential step across a resistor is nega- Fig. 4.6a). Along with the capacitor C, we
tive (positive) if your loop path through the introduce a switch, initially open. The break
resistor is with (against) your guessed cur- in the circuit means that no charge will flow,
rent. so I(t < 0) = 0. With an initial charge ±Q0
We have three equations in three un- on the plates as shown, the top plate has,
knowns – so solve them! Note that the from the definition of capacitance, a poten-
equation from N2 is the same as that from tial V0 = Q0 /C above the potential of the
N1. Similarly, the equation for a loop all bottom plate. This means that, when the
the way around the outside of the circuit switch is closed at t = 0, current will flow
would be a combination of L1 and L2. I get through the resistor, and the capacitor will
I1,2,3 = 18
23
, ≠ 25 , 7 A, where the signs indi-
23 23
begin to discharge. We can treat this analyt-
cate that my guessed directions for I1 and I3 ically using our Loop Rule. We set this up

Copyright 2024 Peter D. Meyers 42

Princeton University Physics 104 Spring 2024

in Fig. 4.6b), which shows the situation at The charge on the cap decays exponentially
an arbitrary time t > 0, choosing the direc- with a time constant · © RC, the time13 it
tions of the current and our traversal of the takes for the charge to decay to 1/e ¥ 1/3
loop arbitrarily. Starting from the lower-left of its original value. A typical resistor value
corner, the Loop Rule gives is 100 , and a typical cap is 1 µF, giving
Q(t) · = 100 µs.
+ ≠ I(t)R = 0. (4.12)
C Something to ponder: At the beginning
We make progress by realizing that Q(t) and of this chapter, we found that even in a
I(t) are clearly related: I in our chosen di- very large current, the electrons making it
rection means that Q is decreasing.12 In up drifted very slowly in the wire. It would
fact, I(t) = ≠dQ(t)/dt, with the minus sign take minutes for an electron to go around a
to give the “Q decreasing” part when +I is modest-sized loop. And yet we found the
in the direction shown by the arrow. (Had current, treated as being the same every-
we picked I pointing the other way, then we where in the circuit, is a maximum imme-
would have I(t) = +dQ(t)/dt. There is no diately after the switch is closed. And, with
rule beyond that.) We see that our Loop our typical R and C, the capacitor is dis-
Rule has given us a differential equation charged in a fraction of a second. How can
this be?
Q(t) dQ(t)
+R = 0, or
C dt !
dQ(t) Q(t)
=≠ . (4.13) ℰ "
dt RC
This differential equation can be solved
by just integrating it, if you know the trick. Figure 4.7: Initially, the charge on the capacitor
Here it is: we multiply through by dt and is zero. It charges when the switch is closed.
divide through by Q(t), then integrate both
sides: Figure 4.7 shows a different RC configu-
⁄ Q(t) ⁄ t
dQ(t) dt dQÕ (t) dtÕ ration. In this case, we start with Q = 0 and
=≠ ∆ = ≠ ,
Q(t) RC Q(0) QÕ (t) 0 RC charge the capacitor with a battery through
-Q(t)
- tÕ --t Q(t) t the resistor. Before starting a detailed analy-
ln QÕ - =≠ - ∆ ln =≠ sis, we can see some things immediately. See
Q0 RC 0 Q0 RC
if you can make simple arguments for these
Q(t) = Q0 e≠t/RC © Q0 e≠t/· . (4.14)
assertions: 1) A long time14 after the switch
12
The charge on one plate is always equal and op- is closed, the current will be zero. 2) The fi-
posite to the charge on the other plate. It is common nal charge on the capacitor will be Q = CE.
to refer to “the charge Q on the capacitor,” meaning
13
one plate has +Q and the other plate ≠Q. The total Show that RC has units of seconds.
14
charge on the cap is always zero. Compared to RC.

Copyright 2024 Peter D. Meyers 43

Princeton University Physics 104 Spring 2024

Choosing positive current to be clockwise components is “ground”, the term for a ref-
and going clockwise around the loop, we get erence point held at the potential of, well,
the ground.16 In our exercises, it is just a
E ≠ IR ≠ Q/C = 0. (4.15) reference that we can call V = 0, but in real
circuits it is a safety feature. You will see
With our choice of current and traversal di- more about this in lab.
rections, I = +dQ/dt. The resulting differ-
ential equation,
Symbol Name Voltage
dQ E Q
= ≠ , (4.16)
dt R RC wire ∆" ≈ 0

can be solved by the same method used in % = 0 or

Eq. 4.14. Try it! The tactical goal is to have switch
∆" = 0
dt on one side and dQ/(something with Q)
on the other.15 With practice (which you battery ∆" = ℰ
will get), you should arrive at ℰ
1 2 +"
Q(t) = Q0 1 ≠ e≠t/· , (4.17) capacitor ∆" = −)/+
#
with · = RC and Q0 = CE. The functions
" resistor ∆" = −%,
for charging and discharging are shown in
Fig. 4.8.
ground "=0

Figure 4.9: Table of components. For each, the

entry in the loop equation is given, assuming
(Table dragged o
positive current flows to the right and the com-
ponent is being traversed left-to-right.

Figure 4.8: Q(t) for the discharging and charg-

ing RC circuits.

We end with a table summarizing our cir-

cuit components so far, Fig. 4.9. One of these
15
I rearranged the equation to get EC≠Q
dQ
= dt
RC
16
Buildings generally have a conducting pipe
and integrated, noting that Q(t = 0) = 0. pounded into the ground for this purpose.

Copyright 2024 Peter D. Meyers 44

Chapter 5: Magnetic Forces and Fields
5.1 Magnetic force if B
˛ is chosen in the direction shown in
Fig. 5.1e), that is, if B
˛ points into the page.1
Consider the situation shown in Fig. 5.1a): a In lecture, you will see the simple experi-
charge q at rest. We observe that the charge ments that show that currents, that is, mov-
remains at rest, and conclude that the to- ing charges, both react to and cause mag-
tal force on q is zero. If this is true with no netic fields, just as charges themselves, mov-
mechanical forces applied to hold q in place ing or not, both react to and cause electric
we can conclude that E ˛ = 0 in this region
fields.
of space. However, as we have now moved
beyond electrostatics, in b) we give q an ini- The magnitude of the magnetic field has
tial velocity and note its path is not straight: an SI unit called the tesla (T), and we see
in this region of space, there is a force on from Eq. 5.1 that 1 T = 1 N·s/(C·m). To
the charge when it is moving, but not when exercise your understanding of Eq. 5.1, con-
it is at rest, a force proportional to v. We vince yourself that a charge q moving with
attribute this to a new field, the magnetic speed v perpendicular to a magnetic field B˛
field, which is given the symbol B. ˛ The ob- with no other forces on it (as in Fig. 5.1b))
servations shown in parts c) and d) of the will move in uniform circular motion. Then
figure, where we try different directions for find the radius of the circle and show that
˛v in the same B, ˛ indicate that the force is the period of the uniform circular motion is
not in the direction of B.˛ Using the right- independent of v.2
hand rule (RHR) to evaluate the cross prod- 1
If your cross products and/or RHR are rusty,
uct, convince yourself that Figs. 5.1b)-d) are work on them in office hours.
consistent with the rule 2
This is one of the key realizations in the inven-
tion of the cyclotron, an early form of particle accel-
F
˛ = q ˛v ◊ B
˛ (5.1) erator. Look up how it works.

i al
Eha
b
ETI
4
E E
d e
Figure 5.1: A charge q in a region with E
˛ = 0. The charge is at rest in a) and moving with velocity
˛v in b)-e). We infer that there is a magnetic field B,
˛ pointing into the page.

45
Princeton University Physics 104 Spring 2024

We can rewrite Eq. 5.1 to apply to a cur- uniform B,˛ as shown in Fig. 5.2. Using
rent in a wire. We will find the force on a Eq. 5.3, confirm the directions of the forces
short segment of the wire with length d¸ and on the left and right sides, and that the forces
cross sectional area A, with the axis of the on the other two sides are zero.3 Just from
wire (that is, the direction of current in the the picture, you can see that the result of
segment) ŝ. Referring to Fig. 4.1 and Eq. 4.1 having the current loop in a magnetic field is
and using the same notation, we can write that, in this orientation, the magnetic force
the amount of moving charge in the segment will make the loop rotate about the dashed
as qnA d¸, with n the number density of free line. This suggests that we find the torque
charges q. This charge is moving with aver- about that axis, which is
age velocity vd ŝ. Using Eq. 5.1, in a mag- ÿ 3 4
netic field B,
˛ the force on the segment is ˛· = ˛ = 2 a (IaB)ˆ
˛r ◊ F · = Ia2 B ·ˆ
2
dF = (qnA d¸)vd ŝ ◊ B.
˛ ˛ Defining d˛¸ © ŝd¸
=µ˛ ◊ B,
˛ with µ ˛ © I A,
˛ (5.4)
and rearranging,
dF
˛ = qnvd A d˛¸ ◊ B
˛ = I d˛¸ ◊ B,
˛ (5.2) with ·ˆ in the direction of ˛· in the figure and
with A˛ the area of the loop, assigned a di-
where we have used Eq. 4.1 in the last step. rection normal to the loop. This result turns
To find the force on a wire, you would inte- out to be more general than our derivation-
grate this over the (possibly curvy) wire. For by-example would suggest. µ ˛ , the product
a straight wire in a uniform magnetic field, of the loop’s current and area, is called the
magnetic moment of the loop.
F
˛ = IL
˛ ◊ B,
˛ (5.3)
This trick turns out to be a Big Deal. As
with L˛ the length (with direction) of the you’ll see in lecture, with a little bit of fussing
wire. with details, we will turn this into an electric
motor.4
F E Ta
I I
5.2 Making a emagnetic
e e

FA a s
a
ta field: Biot-Savart
a
a'iz
Ty
a
I

7 F Electric7charges make electric fields. Moving

electric charges make magnetic fields.5 There
Figure 5.2: A uniform magnetic field applies a
3
torque ˛· to a current loop. Note that the forces on the two leads bringing
current to and from the loop cancel each other.
4
Teslas indeed!
As an example with widespread applica- 5
You might think that magnetic charges would
tion, consider the square current loop in a make magnetic fields. And they would,
I if they ex-
e
I
Copyright 2024 Peter D. Meyers 46
f I B
F o 7
Princeton University Physics 104 Spring 2024

is a formula for the magnetic field produced Figure 5.3 shows a long, thin, straight
by a charge q moving at velocity ˛v, but, in wire (we will consider it infinite) along the
most applications, the moving charges are x-axis, carrying a current I. We want to
currents in wires, and the most useful rela- find the magnetic field due to this current a
tion is the Biot-Savart Law which gives the distance y from the wire. An arbitrary seg-
contribution dB ˛ to the magnetic field at a ment d˛¸ is shown, along with some quantities
point located a distance ˛r from a segment of needed to evaluate dB ˛ using Eq. 5.5. First,
wire d˛¸ carrying a current I using the RHR, convince yourself that dB ˛
from the segment shown points out of the
µ0 I ˛ page, which we will call the z direction, and
dB
˛ = d¸ ◊ r̂. (5.5)
4fi r2 d˛¸ ◊ r̂ = k̂dx sin ◊.6 Then, convince yourself
that this is true for every segment along the
Note that, like Coulomb’s Law, the contri- wire – we just have to add the dBz ’s. Biot-
bution to the field falls off as the inverse- Savart (Eq. 5.5) gives
squared of the distance from the source.
Biot-Savart contains a new constant, µ0 , µ0 I µ0 I
dBz = dx sin ◊ = sin ◊ d◊, (5.6)
which in SI units is exactly 4fi◊10≠7 T·m/A. 4fir 2 4fiy
The invitation to integrate (“dB”)
˛ is not just where, since rewriting in terms of the angle
mathematical, it is realistic: there is no “seg- ◊ is a good tactical move, we have used the
ment of current” that is physically meaning- hard-won relations in Eq. 1.16 with R æ r
ful on its own. Such segments are parts of in the last step. With this move, the integral
circuits. is easy. Integrating from the (infinitely) far
left to far right, ◊ goes from 0 to fi:

5.2.1 B
˛ from current in a long µ0 I ⁄ fi µ0 I 1 -fi 2
Bz = sin ◊ d◊ = ≠ cos ◊--
4fiy 0 4fiy 0
straight wire
µ0 I
= . (5.7)
2fiy
There is nothing special about the point we
⨀ "# picked, so we conclude that the magnetic
!
y field a distance r from a long, straight wire
I (' q carrying a current I is
x
µ0 I
dℓ B(r) = , (5.8)
Figure 5.3: Magnetic field from a long, straight 2fir
wire with current I. pointing around the wire in a direction given
by a new RHR: point the thumb of your right
isted. So far, after a century or so of searching, it
6
appears that they do not. Verify this!

Copyright 2024 Peter D. Meyers 47

dq
Rdq y
Princeton University Physics 104 Spring 2024

hand along the wire in the direction of the loop, the total B ˛ must point up and nor-
current with your fingers curling around the mal to the loop. Calculating B ˛ everywhere
wire, and B
˛ is in the direction your fingers is quite difficult, but we can easily find B
˛
point. Try this out on Fig. 5.4. along the axis of the loop from Biot-Savart.

q
!" !"′

' z '
! !
⨀I &'
q
&' ′
!$ ⨀ a ⨂!$′
I
Figure 5.6: Magnetic field from a loop, applying
Biot-Savart.

Figure 5.4: Magnetic field from long straight

wire.
Choosing a segment d˛¸ as shown in
Fig. 5.6, we note that
- r̂ is- always perpendic-
ular to d˛¸ so that --d˛¸ ◊ r̂-- = d¸, and that dB
˛
has the direction shown. (Check this with
5.2.2 B
˛ from a current loop the RHR.) Symmetry helps us here: the seg-
ment d˛¸Õ gives dB˛ Õ , cancelling any horizontal
component of the field. We thus need calcu-
e a late only Bz = B cos ◊. Using Eq. 5.5,
a Ni
i j dBz =
µ0 I
cos ◊ d¸
2
4fi r 2
t B Bz =
µ0 I
cos ◊
⁄
d¸ =
µ0 I
cos ◊ 2fia
Figure 5.5: Magnetic field from a loop of current 4fir2 4fir2
I. µ0 I a µ0 Ia2
= 2 a= , (5.9)
2r r 2 (z 2 + a2 )3/2
Figure 5.5 shows the simplest circuit, a
s
current loop. We already know something where d¸ is the circumference of the loop.
about B ˛ from the loop: very close to the
We can look at the behavior of this mag-
wire, the magnetic field must look like the
netic field as we move far from the loop (still
field of a straight wire, as indicated by the
along the axis) so that z ∫ a. Then
green field lines. Some more qualitative
thinking about Biot-Savart should convince µ0 Ia2
you that, in the plane of the loop inside the Bz ≠æ . (5.10)
z∫a 2z 3

Copyright 2024 Peter D. Meyers 48

Princeton University Physics 104 Spring 2024

I⨀
We see that far from the loop, B ˛ falls off like a
*
r *#
*"
1/z . This is the behavior we saw in Eq. 1.9
3

for the electric field in the “electric dipole” 0 #′ # %&! L

setup of Fig. 1.5, and, in fact, the magnetic %#′
I⨂
field far from the loop looks identical to the
electric field of the electric dipole in Fig. 1.8, Figure 5.8: A solenoid of length L and radius a
rotated 90¶ to have the x axis along the axis with n turns/m carrying a current I. To super-
of the loop. A loop of current is a magnetic pose the B’s
˛ from the loops, find the field at z
dipole.7 due to the loops in dz Õ at z Õ .
With a current of fixed magnitude, you
can make a bigger magnetic field from a loop
by winding multiple turns of wire. If there segment of the solenoid of length dz Õ carries a
4

are N turns, and the resulting coil is com- current nI dz Õ . We first find the contribution
pact, the magnetic field is just N times that to Bz at position z due to such a segment at
of a single turn. zÕ.

5.2.3 A solenoid
µ0 nIa2 dz Õ
dBz = 1 23/2 , (5.11)
I
I
I I 2 (z ≠ z Õ )2 + a2

Figure 5.7: A solenoid, a continuous coil, treated

as a stack of loops. where we’ve made the appropriate changes to
Eq. 5.9 to match our situation: Bz æ dBz ,
A solenoid is a coil whose length is not z æ (z ≠ z Õ ), I æ nI dz Õ . Now you “just
negligible. If there are many turns closely integrate from z Õ = 0 to L”, where the scare
packed, we can treat this as a stack of circu- quotes indicate we’re in for some rough go-
ing. At this point, just looking up the in-
3
lar loops, each with the same current I – see
Fig. 5.7. We can find B ˛ along the axis of the tegral would be fine; the approach shown in
solenoid by superposing the fields of the indi- Eq. 5.12, is here for completeness – follow
vidual loops using Eq. 5.9. Our setup for su- it through, but don’t worry about having
perposing the loops is shown in cross-section to do it again. First, back up one step in
Fig. 5.8. If there are n loops per meter, each Eq. 5.9, recognizing that the denominator in
7
Eq. 5.11 is still just 2r3 . Then we transform
In the absence of magnetic charges, called “mag-
netic monopoles,” the magnetic dipole is the build- from z Õ as an integration variable to ◊ as de-
ing block of magnetic fields. Even a bar magnet is a fined in the figure, using z Õ = z ≠ a/ tan ◊,
magnetic dipole. dz Õ = a d◊/ sin2 ◊, and r = a/ sin ◊. “Then,

Copyright 2024 Peter D. Meyers 49

Princeton University Physics 104 Spring 2024

5.3 What about magnets?

It turns out that actual magnets, solid ob-
jects that produce a magnetic field, are much
harder to understand than field-producing
currents. A piece of iron can be magnetized
by placing it in a strong-enough magnetic
Figure 5.9: A solenoid of length L = 6 and ra- field. It then permanently produces a mag-
dius a = 1. netic field that is dipole-like, very similar to
a loop or short solenoid. (See Fig. 5.10.)

it’s easy.”

mG
Itm
µ0 nIa2 dz Õ µ0 nI
dBz = = sin ◊ d◊,
2r 3 2
µ0 nI ⁄ ◊2
Bz = sin ◊ d◊
2 ◊1
µ0 nI
= (cos ◊1 ≠ cos ◊2 ) . (5.12)
2
In any given case, cos ◊1,2 can be read off the Figure 5.10: A bar magnet (bottom) makes a
WRONG RIGHT
picture. The formula works anywhere along dipole magnetic field like a short solenoid (top).
the axis, inside or outside of the solenoid.
Figure 5.9 shows the magnetic field along the The end of a magnet out from which B ˛
axis of a solenoid with L = 6a. points is called the north pole of the mag-
net. If you hang the magnet from a string at
A couple cases of interest:
its center, it will turn so that its north pole
• At the center of the solenoid, cos ◊2 = points North.8 We will use magnets in lec-
≠ cos ◊1 , so ture, but this is all you need to know about
them.
Bcent = µ0 nI cos ◊1cent
L/2
= µ0 nI Ò
(L/2)2 + a2
. (5.13) 5.4 Ampère’s Law

• For an infinitely long solenoid, ◊1,2 = Ampère’s Law is another relation between
0, fi and currents and the magnetic field, consistent
8
There is a conundrum here, as you will see in
B(L æ Œ) = µ0 nI. (5.14) lecture.

Copyright 2024 Peter D. Meyers 50

Princeton University Physics 104 Spring 2024

with Biot-Savart, but useful in different

cases. Ampère’s Law, which applies only to
steady (unchanging) currents, and thus mag-
netostatics, states that i
j
B
˛ · d˛¸ = µ0 Ithru . (5.15)
C
Figure 5.12: Ampère’s Law applied to a long,
straight wire with current running out of the
The integral is over any closed path C, and page.
Ithru is the total current flowing through the
loop formed by the path. A general case
is shown in Fig. 5.11. The angle a current is easy and, using Eq. 5.8, we get
makes with the plane of the loop makes no
j j j
difference, but there is a rule for the sign of B
˛ · d˛¸ = B(r) d¸ = B(r) d¸
a current, a RHR: With the fingers of your C C C
right hand curling in the direction you go =
µ0 I
2fir = µ0 I, (5.16)
around the path (counter-clockwise in the 2fir
figure), your thumb points in the direction
assigned as a positive current. Hence, in which Prof. Ampère asserts, correctly, is
Fig. 5.11, Ithru = I1 ≠ I2 + I3 ≠ I4 . equal to µ0 Ithru . Prof. Ampère also asserts
that the line integral around path C Õ gives
the same answer.9

5.4.1 Finding B
˛ using
Ampère’s Law
Figure 5.11: Ampère’s Law setup.
For us, the immediate utility of Ampère’s
Law is that, in cases of sufficient symmetry,
We can verify Ampère’s Law with a case we can use it to find B.
˛ It thus plays the role
we know: the magnetic field around a long, for B that Gauss’s Law plays for E.
˛ ˛
straight wire carrying a current I, Eq. 5.8.
We choose an easy-to-analyze path C, a cir- 9
See if you can convince yourself that this is true.
cle centered on the wire, which we will follow Try an argument that breaks C Õ into short segments,
in the same direction B˛ points. Because ev- each of which is radial or a circular arc centered on
erywhere along C, B Î d˛¸, the dot product
˛ the wire, and then uses the fact that B Ã 1/r.

Copyright 2024 Peter D. Meyers 51

Princeton University Physics 104 Spring 2024

Example: a long, straight wire with the current uniformly distributed over
its cross-section.10 First think of how, in
As an example, let’s find B ˛ for a long, principle, you might set this up with Biot-
straight wire carrying a current I again from Savart. Now look step-by-step through our
scratch, this time using Ampère’s Law, aided approach in Eq. 5.17 using Ampère’s Law,
by symmetry and a bit of Biot-Savart. Just asking yourself what, if anything changes.
the d˛¸ ◊ r̂ structure of Biot-Savart tells us The answer: nothing changes. We have just
that all contributions to B,
˛ and thus B ˛ it- used Ampère’s Law to show that our re-
self, point azimuthally around the wire, as sult in Eq. 5.8 is exactly valid outside an
shown in Fig. 5.12. Symmetry tells us that azimuthally-symmetric current of any width.
B
˛ has the same magnitude at all points with
We’re not done. Now consider a point in-
the same r. These together suggest that to
side the wire, that is, at r < R. Again look
find B(r) we choose a circle of radius r, cen-
at the steps in Eq. 5.17, now with C a cir-
tered on the wire, as our path. Then
cle centered on the axis of the wire that is
j j j inside the wire. All the steps are still valid.
B
˛ · d˛¸ = B(r) d¸ = B(r) d¸ One thing does change: now Ithru ”= I, it
C C C
= B(r)2fir = µ0 Ithru , (5.17) is only the current within C. For a current
uniformly distributed across the area of the
where our choice of C allows us to use the wire, Ithru (r) = I(r2 /R2 ), so altogether we
same reasoning as in the first line of Eq. 5.16, have
but with B(r) still unknown, and the last Y
step is the application of Ampère’s Law. The µ0 Ithru (r) ] µ2fir
0I
,for r > R;
B(r) = = µ0 Ir
last line gives us the same expression we got 2fir [
2fiR2
for r Æ R.
for B(r), with much more effort, from Biot- (5.18)
Savart. In fact, we will assume that, after See Fig. 5.13 to understand the geometry.
some practice, you will be able to immedi-
ately derive B˛ for a long, straight wire using
Ampère’s Law, showing the reasoning. R
To better see the power of Ampère’s Law, r
C
I, uniform
we can easily handle a new case that would !
throughout wire
be quite difficult with Biot-Savart. So far, we
have treated our wire as having a negligible Figure 5.13: Ampère’s Law applied inside a wire
width. Now consider what would change in carrying a uniform current.
the above argument if the wire had a circu-
lar cross section and a non-negligible diam-
eter. Say, for example, the wire in Fig. 5.12 10
In fact, I can be non-uniform, as long as it is
had a radius R not much smaller than r, azimuthally symmetric.

Copyright 2024 Peter D. Meyers 526

Princeton University Physics 104 Spring 2024

Example: a circular loop any calculation. An initial observation: since

the solenoid is infinite, B˛ cannot vary in
#$%&' the direction parallel to the solenoid’s axis,
which we will take to be the z axis. It must
also be azimuthally symmetric, having the
a !ℓ
same magnitude for all points the same dis-
I tance from the axis.
C y
a B z

Figure 5.14: Ampère’s Law applied to a circular

loop of current.

Figure 5.14 shows a setup for finding the b I

magnetic field at the center of a circular loop
of radius a carrying a current I. We use a
path C, also a circle with radius a, and apply 5
Ampère’s Law thusly: 7
d
j j
B
˛ · d˛¸ = µ0 Ithru = B d¸ = B2fia
C C

=∆ Bcent =
µ0 I
. (5.19) d B
2fia
r
Take a look at our Biot-Savart result for
this case, Eq. 5.9 with z = 0 – the result
in Eq. 5.19 is wrong. Critique in detail each
step in Eq. 5.19. What went wrong? (I’ll Figure 5.15: Inferring the direction of B
˛ for an
infinite solenoid.
start: the first step is correct – it is Ampère’s
Law.)
What about the direction of the field?
From our solution to one loop (Sec. 5.2.2),
Example: a long solenoid we know that B ˛ on the axis points along the
axis. What about off the axis? In Fig. 5.15a),
A long (read: infinite) solenoid is a good we explore the possibility that B˛ in the plane
example for practicing the symmetry argu- of the figure has a y component. In b) we
ments crucial to the use of Ampère’s Law to have flipped a) end-to-end, resulting in B ˛ as
find B.
˛
shown. However, as you can see, flipping the
As with using Gauss’s Law to find E, ˛ we solenoid in this fashion is identical to simply
can often infer features of the field without reversing the current in the loops. This must

Copyright 2024 Peter D. Meyers 53

Princeton University Physics 104 Spring 2024

reverse B
˛ everywhere, as shown in c). b) and since C1 encloses no current. Hence B2 = B1
c) cannot both be true: we conclude that B ˛ and the magnetic field inside an infinite
has no y component. solenoid is constant in magnitude and
What about an x component, shown with direction. The same reasoning applied to
a question mark in c)? Part d) of the fig- C2 says the field outside the solenoid is also
ure shows this hypothetical case viewing the uniform.11
solenoid from the end. Rotating about the We can use path C3 to link the fields in-
axis of the solenoid can change nothing, so side and outside the solenoid. Ampère’s Law
we infer that if B˛ is as shown, then the field gives
must curl around as indicated by the dashed j
arrows. Ampère’s Law applied to the dashed B
˛ · d˛¸ = B1 ¸3 + 0 + B3 ¸3 + 0
circle tells us that 2firB = µ0 Ithru = 0, since
C3

there is no current through the circle. Hence = (B1 + B3 )¸3 = µ0 Ithru = µ0 nI¸3
B
˛ has no x component. Our conclusion: B ˛ ∆ B3 = µ0 nI ≠ B1 . (5.21)
is parallel to the axis of the solenoid
What’s interesting about this: in Eq. 5.14 we
everywhere, inside and outside of the
found that, for an infinite solenoid, the field
solenoid.
on the axis is B1 = µ0 nI, so the magnetic
l field outside an infinite solenoid is zero,
C2 sa 4 or
I BJ Y
S3 ]µ
0 nI k̂, inside;
C SYL
B Cz B
˛ Œ solenoid = (5.22)
[0 outside.
l lz
Figure 5.16: Ampère’s Law applied to an infinite This result is useful in the real world: a
solenoid. long solenoid can be used to make a region
of space that has a uniform magnetic field.
Figure 5.16 shows how we can exploit It is useful in PHY104 because it is a situ-
our understanding. We explore the field via ation in which the magnetic field is uniform
Ampère’s Law applied to closed paths with and known – the starting point for a shock-
legs either parallel to B
˛ or perpendicular to ing number of problems. For this reason,
it. Since B does not vary with z, going I give you, in writing, this special dispen-
around path C1 , starting at the lower-left sation: when faced with a long solenoid in
corner, gives PHY104, you may simply assert that 1) B ˛ is
j 11
I have drawn the hypothetical B ˛ outside the
B
˛ · d˛¸ = B1 ¸1 + 0 ≠ B2 ¸1 + 0
solenoid pointing opposite to that inside only be-
C1
cause that’s how it works for a single loop. As you’ll
= (B1 ≠ B2 )¸1 = µ0 Ithru = 0, (5.20) see, this doesn’t matter.

Copyright 2024 Peter D. Meyers 54

Princeton University Physics 104 Spring 2024

parallel to the axis, and 2) B = 0 outside, far (about halfway!) in this compact form:
and then use Ampère’s Law to find B ˛ any-
where inside using a path like C3 in Fig. 5.16. j Electrostatics Magnetostatics
j
This will give you full credit, while simply re- E
˛ · dA
˛ = Qin /‘0 B
˛ · dA
˛ =0
membering the result will give you none. S j j S
E
˛ · d˛¸ = 0 B
˛ · d˛¸ = µ0 Ithru
C C
(5.25)
5.5 Filling in a few gaps There is an intriguing pattern here, which we
will flesh out in the coming weeks.
We have seen a parallel between Gauss’s Law
for electrostatics and Ampère’s Law for mag-
netostatics in how, in cases of sufficient sym-
metry, we can use them to find their respec-
tive electric and magnetic fields. However,
they are not parallel mathematically. There
is, in fact, a Gauss’s Law for magnetism.
Because there are no magnetic charges,12
Gauss’s Law for magnetism is simply
j
B
˛ · dA
˛ = 0. (5.23)
S

We have already met in passing (Sec. 3.1)

“Ampère’s Law for electrostatics” (though it
is never called this). Since Coulomb’s Law is
a conservative force, the line integral of E
˛ be-
tween any two points is independent of path.
This means that the line integral around a
closed path is zero,
j
E
˛ · d˛¸ = 0. (5.24)
C

We can summarize a lot of the course so

12
Often referred to as “magnetic monopoles”. The
fundamental producer of magnetism seems to be the
magnetic dipole.

Copyright 2024 Peter D. Meyers 55

Chapter 6: Electromagnetic Induction
Figure 6.1 shows a gallery of observations
that we will do in lecture this week. To-
gether, they broaden our horizons on the be-
havior of electric and magnetic fields, open-

is
ing the way to new technologies and new the-
oretical insights.
In Fig. 6.1a) we have a situation that we
can understand fully with what we already
know. The end of a bar magnet makes a
spreading magnetic field that at a distance
has the now-familiar dipole behavior. A loop
of wire is moved by hand at a velocity ˛v to-
ward the north pole of the magnet. Consider
two positive charges in the wire, one at the
top of the loop and one at the bottom. They
too are moving with velocity ˛v, and in the
field of the bar magnet, each experiences a
force given by Eq. 5.1, F˛ = q ˛v ◊ B.
˛ The di-
rections of these forces, into the page at the
top and out of the page at the bottom, are
shown in the figure. Figure out for yourself
how the force points at other locations in the
loop. You should find that the force points
around the loop. Now, we know that the pos-
itive charges in the wire are not free to move E
relative to the wire. However, some of the my
electrons are free to move, so ’round they go,
out of the page at the top of the loop and into
the page at the bottom. (Convince yourself d
that the direction of this current is the ar-
row shown in the figure, which is meant to
represent the near side of the loop.) By mov-
ing the loop, we’ve made a current. This is
what an EMF does – we’ve induced an EMF
nie
Figure 6.1: A gallery of observations.
without using a battery! (See Sec. 4.2.)

56
Princeton University Physics 104 Spring 2024

Now take a moment to consider what Fig. 6.1c). This is also confirmed experimen-
would happen if the same loop was moved tally. Again, we expect the same current in
in a uniform magnetic field, with the same the loop if the coil moves towards it, and
magnitude and direction everywhere. Make again, this is what is observed.
some sketches, try various directions for B.
˛
You should find that in none of these situa-
tions does this force “point(ing) around the 6.1 Faraday’s Law
wire” occur.
Systematic study of experiments like these
In Fig. 6.1b) we have the same magnet (carried out in the 1830’s by Michael Faraday
and loop. In this case, the loop remains sta- in England and Joseph Henry at Princeton1 )
tionary and the magnet moves with velocity led to the conclusion that the crucial feature
˛v toward the loop. The charges in the loop these phenomena have in common is that the
are at rest, so the magnetic force on them is magnetic flux through the loop is changing.
zero. And yet we know what the outcome The magnetic flux through the loop is defined
must be: situation a), in which there is a in parallel to the electric flux in Eq. 2.2 as
current of electrons in the ring, can be trans- ⁄
formed into situation b) by simply chang- „m © B
˛ · dA,
˛ (6.1)
S
ing reference frames (i.e., by walking along
with the moving loop). This cannot change where S is a surface spanning the loop. Fara-
whether there is a current or not. Experi- day’s Law relates the induced EMF around a
ment shows that, indeed, the current in the loop C to the change in the magnetic flux
loop when the magnet moves is the same as through a surface S spanning C:
the current when the loop moves. The only
problem is that we do not (yet) have a phys- d„m d ⁄ ˛
E =≠ =≠ B · dA.
˛ (6.2)
ical explanation for this. dt dt S

The induced current in the loop is then given

We can go further, even without such a by Ohm’s Law: I = E/R, with R the resis-
theory. While there is no magnetic force in tance of the loop.
situation b), the current clearly requires a
This focus on the change in magnetic flux
magnetic field to be present. (Replacing the
suggests that no motion at all is needed
magnet with a moving stick yields no cur-
to induce an EMF. In Fig. 6.1d), the cur-
rent.) This suggests that we should get the
rent in the coil is turned on by closing the
same current in the loop if we replace the bar
switch. This certainly changes the magnetic
magnet with a coil of wire with its own cur-
flux through the loop, and Faraday’s Law
rent, I0 , with the coil and the current cho-
sen to give the same B ˛ at the loop, as in 1
!

Copyright 2024 Peter D. Meyers 57

Princeton University Physics 104 Spring 2024

states that there will be a momentary EMF

and current around the loop as the magnetic increasing J B induced
field turns on. This, too, is observed. E If I da TI DA

E
6.1.1 Lenz’s Law
al b b
The minus sign in Eq. 6.2 comes with a con-
vention: with your right thumb pointing in Figure 6.3: A loop in a uniform but increasing
the direction of positive flux, that is the di- magnetic field.
rection you choose for dA ˛ on S, a positive
EMF E will be the direction around the loop
2. By our RHR, a positive EMF E would be
C given by your curling fingers. This is illus-
in the direction indicated by the arrow
trated in Fig. 6.2.
with a + sign on the loop.
3. Our choice of dA ˛ to the right also means
+ that flux produced by the uniform B ˛ will
S
be positive, since B·dA is positive every-
˛ ˛
!" where over a surface spanning the loop.
4. Since the flux is positive and increasing
+ in magnitude, d„m > 0, and the minus
C sign in Faraday’s Law means that E is
Figure 6.2: The right-hand rule for Faraday’s actually negative.
Law. The choice of the + direction of dA
˛ deter- 5. We conclude
4 that the induced EMF and
mines the + direction for E and I. the current in the loop are negative,
meaning, from our sign convention, that
they go in the direction shown.
With our sign convention in place, we still
have to confront that minus sign in Faraday’s You would be wise to take a moment to go
Law. Let’s look at a simple case, shown in through this again yourself, and then do the
Fig. 6.3. We have a loop of wire, as before. It case where you pick dA ˛ pointing to the left
is in a uniform B,
˛ pointing to the right. This (which should give the same direction for
magnetic field is increasing in magnitude.2 E)3 , and then do the case with the origi-
nal dA˛ but a decreasing magnetic field, still
Taking it step-by-step: pointing to the right (which should give the
1. In Fig. 6.3a), we have already chosen dA
˛ opposite direction for E).
(arbitrarily) to point to the right.
The current induced in the loop produces
2
We can imagine that the loop is in a big solenoid,
3
and that the current I0 in the solenoid is increasing. Why?

Copyright 2024 Peter D. Meyers 58

Princeton University Physics 104 Spring 2024

its own magnetic field B ˛ Õ – I sketched this 6.1.2 Examples

field in Fig. 5.5 and derived its magnitude
along the axis of the loop in Eq. 5.9. In Induced current
Fig. 6.3b), I add this field in purple and note
that it points to the left through the loop.
There is a pattern here, but it isn’t the ob-
vious one (i.e., “The induced B ˛ Õ is in the opposite
'
˛ 4 ). If you (wisely)
direction to the original B”
did the exercises suggested in the previous
paragraph, you found that if the original B ˛
( &
pointed to the right but was decreasing, the
induced current in the loop went in the op-
posite direction. In that case, the induced B ˛Õ " # = %#
is in the same direction as the original B. ˛ Figure 6.4: A loop near a long wire with a time-
dependent current.
The pattern here is important enough to
have its own name. Lenz’s Law states The Figure 6.4 shows a long wire carrying a
induced EMF is in a direction such that time dependent current I(t) = –t, with –
the resulting current would produce a a positive constant with dimensions A/s. A
new magnetic field that opposes the ⨂*. *) resistance
rectangular loop of wire with ⨂ +()) R is
change in flux. Practically, Lenz’s Law is a distance s '
from the wire. Find the direction
a much more reliable way to determine the and magnitude of the induced current I Õ in
direction of induced currents than trying to the loop. )
( &
wrestle with the RHR, sign convention, and We start by trying to make sense of the
minus sign in Faraday’s Law. However, there situation. Is there a changing „m through
is also something fundamental going on here. the loop? Yes: "the
# long
= %#wire is making B ˛
Consider the case in Fig. 6.3 again. What if that is time-dependent because I is. We can
the induced current I went in the other di- immediately find the direction of the induced
rection? Then the induced B ˛ Õ would point
current with Lenz’s Law in a few quick steps:
in the direction of the increasing B, ˛ further
1) Using the RHR, B˛ at the loop points ITP.5
increasing the flux through the loop, which 2) The current in the wire is increasing, so
would further increase B ˛ Õ ,... As you might
so is the magnitude of B.
˛ 3) Since we have a
expect, this runaway increase in the current magnetic flux into the page and increasing,
from even the smallest external change of B ˛
Lenz’s Law tells us the induced current in
cannot happen. Looking at the material later the loop will make a new B˛ Õ pointing OOP6
in this chapter, you will see that Lenz’s Law
is enforcing conservation of energy. 5
Into the page. This is not a standard TLA
(three-letter acronym), but I like it.
4 6
In small type because it is wrong, so forget it. Out of the page. Also non-standard.

Copyright 2024 Peter D. Meyers 59

Princeton University Physics 104 Spring 2024

to oppose the change in flux. 4) A CCW7 I Õ Then

in the loop makes an OOP B ˛ Õ . We conclude ⁄ ⁄ s+b
that the induced EMF and current will be „m © B
˛ · dA
˛ = B(x)a dx
S s
counter-clockwise. µ0 I(t)a ⁄ s+b dx µ0 I(t)a s+b
A B
= = ln
2fi s x 2fi s
Now we calculate. First, we need the mag- A B
netic field from the long wire. You should be d„m µ0 a s + b dI
E = (≠) = ln
able to get this using Ampère’s Law in less dt 2fi s dt
A B
than a minute. The field a distance x from µ0 a s+b
= ln –
the wire (Eq. 5.8) is B(x) = µ0 I/(2fix), and 2fi s
we already determined that it points ITP in E µ a
A
s + b
B
0
the loop. Since B is not uniform in the loop, IÕ = = ln – (CCW).
R 2fiR s
we will have to integrate to get the flux, per
(6.3)
the definition Eq. 6.1. Try setting it up be-
'
fore you continue reading. In the third line, I’ve placed the minus sign
from Faraday’s Law in parentheses to show
We take advantage of the fact that B ˛ that 1) I haven’t forgotten it, but 2) we’ve
( &
is independent of the coordinate parallel to already found the direction of E and I Õ from
the wire to set up the integral as shown in Lenz’s Law, so we can ignore it.
Fig. 6.5, with dA
˛ = a dx pointing into the
" # = %#
page. Pulling a loop from a magnet
$ uniform $ = 0
⨀ ⨀ ⨀ ℓ
⨀ ⨀ ⨀
+,
*) ⨂ +()) ⨀" ⨀ ⨀
' ⨂*. ⨀ '⃗
⨀ ⨀ ⨀
⨀ ⨀ ⨀ #
( )
&
Figure 6.6: A loop partially in a magnetic field.
In this example, we again have a rectan-
" # = %# gular loop of wire with resistance R. Now it
Figure 6.5: Setup to find magnetic flux. is partly in a region 5with a uniform magnetic
⨂ field, OOP, as$shown in $ Fig.
= 0 6.6.8 The ques-
ℓ 8
⨀ in B
The abrupt change ˛ is not physical – we
7
Counter-clockwise. This one is standard. imagine it here to make the calculation simpler.
⨀ ⨀ ⨀
⨀$ "
⨀ ⨀Copyright
⨀ 2024 -
Peter D. Meyers
= ./×$ '⃗ 60
⨀ ⨀ ⨀ ./ ⨀
⨀ ⨀ ⨀
⨀ ⨀ ⨀
 ⨀" ⨀ ⨀ ⨀ '⃗
⨀ ⨀ ⨀
⨀ ⨀ ⨀ #

Princeton University Physics 104 Spring 2024

⨂ must you $ $=0

tion: with what force to the right
pull on the loop to have it moveℓat constant ⨀
velocity ˛v? ⨀ ⨀ ⨀
There is a lot hidden in ⨀$ "
⨀ this
⨀ innocent-
⨀ - = ./×$ '⃗
sounding question. Let’s think it through.
⨀ ⨀ ⨀not a
A force should give an acceleration, ./ ⨀
constant velocity. This implies⨀ ⨀that⨀there
must be another force on the loop to the left. Figure 6.7: Finding the force on the loop.
⨀ ⨀ ⨀
What could cause this force? With the loop
entering a region with B ˛ = 0, the magnetic
is given by Eq. 5.3:
flux through the loop is changing, so there
will be a current induced in the loop. Part of F
˛ = IL ˛ = IwB = w2 B 2 v/R
˛ ◊B (6.5)
that current is still in the region with B
˛ ”= 0,
so there could be a force on the current as pointing to the left. (We even see Lenz’s Law
discussed in Sec. 5.1. With that in mind, we in this: the force is literally “opposing the
can pull the pieces together. change in flux”!)
We will first find the induced current in The answer to our original question is
the loop, I, using the methods of the pre- then: to move the loop to the right with
vious example. The flux through the loop constant speed v, we must apply a force
points out the page and is decreasing as the F = w2 B 2 v/R to the right. Can you an-
loop moves into the region with zero field. swer this question: what happens if you stop
Lenz’s Law then tells us that I will be in pulling? Does the loop get pulled back into
the direction to oppose this by making more the magnetic field?
flux OOP. The current to do this is CCW.
(Check this yourself.) In the calculation of
I, the uniform field makes the flux integral 6.2 Changing B
˛ makes E
˛
easy: „m = Bw(¸ ≠ x), where w(¸ ≠ x) is the
area of the loop remaining in the field region. Faraday’s Law, Eq. 6.2, is the key to a wide
Then the current is CCW, with magnitude range of applications that we will barely
scratch the surface of in the next couple
E 1 d„m Bw dx Bwv weeks. Now, however, we turn to a more
I= = (≠) = = .
R R dt R dt R abstract view, an important step towards
(6.4)
a complete picture of Electricity and Mag-
Part of this current is still in the magnetic netism. We start by reconsidering the situ-
field – Fig. 6.7 shows the situation. With ation in Fig. 6.3a), a loop in a uniform, in-
the forces on the top and bottom segments creasing magnetic field. The induced EMF
cancelling, the net force of B ˛ on the current can make the electrons in the loop, initially

Copyright 2024 Peter D. Meyers 61

 Princeton University Physics 104 Spring 2024

at rest, start moving. We do know of another zero. (Recall Eq. 5.24. This is what makes
way to make charges at rest move, without the Coulomb force a conservative force.) You
a battery or a mechanical force: an electric cannot make an electric field like the one in
field. Fig. 6.8 with any arrangement of stationary
We can view the induced EMF, which, un- charges.
like a battery, is distributed around the loop,
We can now extend Faraday’s Law to read
as the result of a new kind of electric field, in-
duced by the changing magnetic flux. In the
case of the loop in Fig. 6.3a), this E˛ is illus-
j ⁄
trated in Fig. 6.8a). We know that EMF has ˛ · d˛¸ = E = ≠ d„m = ≠ d
E B
˛ · dA,
˛
the same units as electric potential – this sug- C dt dt S
gests9 that the induced EMF around a curve (6.7)
C is j or, simply
E= E
˛ · d˛¸. (6.6)
C

In this view, the induced current is driven by j ⁄

the induced electric field pointing around the ˛ · d˛¸ = ≠ d
E B
˛ · dA.
˛ (6.8)
loop. C dt S

⨀ ⨀ ⨀ ⨀ ⨀ ⨀
!" $ !" $
⨀ !# ⨀ ⨀ ⨀ ⨀ ⨀ !# ⨀ ⨀ ⨀ ⨀
This remarkable equation relates E ˛ to B,
˛ or,
more specifically, to the time derivative of B,
˛
⨀ ⨀ ⨀ ⨀ ⨀ ⨀ ⨀ ⨀ ⨀ ⨀ with no reference to EMF or current. In fact,
⨀ ⨀ ⨀ ⨀ ⨀ ⨀ ⨀ ⨀ ⨀ ⨀
the equation suggests that we can simply re-
move the wire from Fig. 6.8a) entirely, as in
a) b)
⨀ ⨀ ⨀ ⨀ ⨀ ⨀ b), making this simply about fields: a chang-
ing B˛ makes E.˛ This whole process is thus
Figure 6.8: E˛ induced by changing B. ˛ Note called electromagnetic induction.
that we have labelled the green OOP vectors as
⨀“dB/dt,”
⨀˛
⨀ which could be an increasing B
˛ OOP, An important detail about Fig. 6.8:
⨀or a decreasing B ITP.
⨀ ˛ Eq. 86.8 gives us only the (line) integral of
⨀
E.
˛ To extract E ˛ itself, as we did in the fig-
ure requires more information, symmetry, for
I described this as a “new kind of electric
example. In Fig. 6.8, we have assumed cylin-
field.” For any electrostatic field, that is, all
drical symmetry about the center of the fig-
those we’ve studied up to now, any integral
ure, allowing us to know, for example, that
over a closed path as in Eq. 6.6 is identically
the magnitude of E˛ was the same everywhere
9
This is a discovery, not a derivation! along the circular path in the integral.

Copyright 2024 Peter D. Meyers 62

 (Table dragged over picture

Princeton University Physics 104 Spring 2024

6.3 Inductance and in- Symbol Name Voltage

ductors &'
L inductor ∆" = −%
&(
The ability to use a changing magnetic field
to manipulate currents is clearly relevant to
our ongoing study of electric circuits. Example: a long solenoid

There aren’t many geometries for which the

6.3.1 Self-inductance inductance is easily calculable. (Can you see
why the inductance of a circular ring would
We introduce a new circuit component, the be hard?) We will find L for a solenoidal coil
inductor, which is a small coil with many of length ¸ and radius a with ¸ ∫ a, using
turns of wire, either flat or solenoidal. The the infinite solenoid field as an approxima-
situation here is different from our study of tion. We found that the magnetic field of an
induction so far. The coil of the inductor infinite solenoid (Eq. 5.22) is uniform inside,
is both the source of the magnetic field and pointing parallel to the axis with magnitude
the path through which the flux changes – B = µ0 nI, where n is the number of turns
a changing current in the inductor produces per unit length. Our inductor has N = n¸
an EMF in the inductor itself. turns. The magnetic flux is thus
In an inductor, the geometry is fixed, and
„m = N fia2 B = n¸fia2 µ0 nI = µ0 fia2 ¸n2 I,
the only change in magnetic flux occurs due
(6.10)
to the changing of B,
˛ which is, in turn, due
where the factor of N appears because the
to a changing current in a coil. In this sit-
flux goes through all N turns. (The turns
uation, the magnetic field, and hence „m , is
are in series, and the EMFs add.) We can
proportional to the current I in the coil. This
then read off the inductance:
means that

dI L = µ0 fia2 ¸n2 . (6.11)

„m = LI; E = ≠L , (6.9)
dt

where the constant L depends on the ge- 6.3.2 An LR circuit

ometry. This constant is called the self-
Figure 6.9 shows an LR circuit. When the
inductance or simply the inductance. The
switch is closed at t = 0, what is the current
SI unit of inductance is the henry,10 with
as a function of t? Before we find I(t), we
1 H = 1 T · m2 /A = 1 V · s/A. We can now
can develop some intuition by asking: what
add a new row to our table of components:
will the final, steady current If be? If the
10
After Princeton’s Joe Henry. current is steady, then dI/dt = 0, and there

Copyright 2024 Peter D. Meyers 63

Princeton University Physics 104 Spring 2024

! I(t) has the same shape as the blue curve for

Q(t) of a charging RC in Fig. 4.8, and, as we
ℰ " figured, I(t ∫ · ) = E/R.

The upshot is that, because of Lenz’s Law,

Figure 6.9: An LR circuit. an inductor opposes the change in current.
In this case, the inductor slows the turn-on
is no induced EMF in the inductor. This of current in the LR circuit, which otherwise
means that the inductor does nothing (it’s would be instantaneous when the switch was
just a wire), and we effectively have just the thrown.
battery and the resistor, a circuit that has a
current If = E/R.
Now we fully analyze the circuit using our
usual rules. The Loop Rule, going clockwise ! !
and taking a clockwise I to be positive, gives
ℰ $" "
dI
E ≠ IR ≠ L = 0. (6.12)
dt
This is a differential equation, in fact the Figure 6.10: Another LR circuit.
same equation (with different symbols) you
got for the charging RC circuit in Fig. 4.7.
We left it as an exercise then, but let’s do it
now.
dI L dI The circuit we just solved is the left side
E ≠ IR = L ∆ dt = ; of Fig. 6.10. The two switches have been in
dt E ≠ IR
⁄ t ⁄ I(t)
L dI the position represented by the solid lines,
dt =
Õ
∆ the left switch closed and the right switch
0 0 E ≠ IR
-t L -I(t) open, for a long time. Our recent calcula-
- -
tÕ - = ≠ ln(E ≠ IR)- ; tion says that the current in the inductor is
0 RA 0
B I0 = E/R; no current flows in the resistor
L E ≠ I(t)R
t = ≠ ln ; on the right. At a (new) t = 0 we simul-
R E
taneously open the left switch and close the
E 1 2 E 1 2
right switch as indicated by the orange ar-
I(t) = 1 ≠ e≠Rt/L © 1 ≠ e≠t/· .
R R rows. What happens? The current I0 must
(6.13)
flow through the right-hand loop. The Loop
I used the substitution u = E ≠ IR to do Rule tells us what happens for t > 0. Going
the integral, and in the last step I’ve identi- around the loop in the direction of the red
fied a time constant · © L/R. This function arrow, with the + direction of the current

Copyright 2024 Peter D. Meyers 64

Princeton University Physics 104 Spring 2024

following the I0 arrow, have been stored in the inductor. We can

compute this energy because we know the
dI
≠L ≠ IR = 0; resistor dissipates energy at a rate given in
dt Sec. 4.3 as P = IV = I 2 R. The energy in
⁄ I(t)
dI R⁄ t Õ
=≠ dt ; the inductor L at t = 0 must have been
I0
A
I B
L 0 ⁄ Œ ⁄ Œ1 22
ln
I(t) R
= ≠ t; U= PR dt = I0 e≠t/· R dt
I0 L ⁄0Œ 0
3 -Π4
·
I02 e≠2t/· R dt I02 R
-
I(t) = I0 e≠Rt/L © I0 e≠t/· . (6.14) = = ≠ e≠2t/· -
0 2 0
3 4
L 1
(Fill in the missing steps.) = I02 R ≠ (0 ≠ 1) = LI02 . (6.15)
2R 2
I plot these results in Fig. 6.11.11 In both
cases, the inductor keeps the current from We conclude that an inductor L carrying cur-
changing abruptly.12 1
rent I has U = LI 2 of stored energy.
2
When we discussed the energy stored in
a capacitor in Sec. 3.6.2 we found that the
stored energy could be considered to reside in
the electric field in and around the capacitor.
A parallel relation exists here, with

1 2 ⁄ 1 2
U = LI = uB dV, with uB = B ,
Figure 6.11: Current ramping up and down in 2 2µ0
an LR circuit. I0 = E/R. (6.16)
where uB is the energy density of a magnetic
field B.

6.3.3 Energy stored in an in-

ductor 6.3.4 Mutual inductance

In the case we just looked at, the right-hand In mutual inductance, we return to a situa-
resistor dissipated energy as heat after the tion like that in Fig. 6.1d): a changing cur-
battery was disconnected. The energy must rent in one component induces an EMF in
another. We take the geometry to be fixed,
11
...which is just Fig. 4.8 with the axis labels so the only change in flux in component 2
changed.
12
The minus sign in ≠L dI/dt is from Lenz’s Law. is due to the changing current in compo-
You might ponder what Eq. 6.14 would give without nent 1. Convince yourself that the flux in
it. component 2 is proportional to the current

Copyright 2024 Peter D. Meyers 65

Princeton University Physics 104 Spring 2024

in component 1, so
dI1
E21 = M21 , (6.17)
dt
where E21 is the EMF induced in component
2 due to the changing current in 1 and M21 is
a constant, in henrys, which depends on the
geometry of the situation.
A very useful property of mutual induc-
tion is the Reciprocity Theorem which states
that for any geometry, M21 = M12 , where
the latter determines the EMF in 1 due to a
changing current in 2. We simply call both
M (hence “mutual”). This means that there
are always two ways to find M , and you get
to choose the easier one.
The most common use of mutual induc-
tion in circuits is in a device with two coils
called a transformer.

Copyright 2024 Peter D. Meyers 66

Chapter 7: AC Circuits
7.1 Oscillating currents ential equation for Simple Harmonic Motion,

d2 Q(t) 1
The addition of the inductor to our circuit 2
=≠ Q(t) ∆
dt LC
repertoire brings with it a new phenomenon, 1
Q(t) = Q0 cos Êt, with Ê = Ô © Ê0 ,
illustrated by the circuit shown in Fig. 7.1. LC
In this circuit, we start with the switch open (7.2)

where we have applied the initial conditions

a) b) to get the amplitude Q0 and cos instead of
+&! +&(#) !(#) some other phase. We Ô have defined the an-
" ! " ! gular frequency Ê0 © 1/ LC (check that the
−&(#) L1
−&! units are 1/s as they should be) for use later.
The current has the same oscillatory be-
havior:
Figure 7.1: An LC circuit. The cap starts out
with charge Q0 . I(t) = ≠
dQ(t)
= Q0 Ê sin Êt = I0 sin Êt.
dt
(7.3)
and the capacitor charged. We would like The current alternates in direction.
to know what happens when the switch is In Ch. 4 we studied circuits driven by
closed at t = 0. The Loop Rule is all we a source of constant EMF – direct current
need here – we will apply it at an arbitrary (DC) circuits. We will now investigate elec-
time t. Choosing positive current as shown tric circuits driven by a sinusoidally oscillat-
by the arrow in Fig. 7.1b), and traversing the ing EMF, called alternating current (AC) cir-
loop clockwise from the lower-left corner, we cuits. You don’t have to look far to find a
have source of such an EMF – this is what you
get when you plug something into a wall
Q(t) dI(t) socket. In the United States, you get an
≠L =0∆ EMF E(t) = 170 cos(377t) V, with t in sec-
C dt
Q(t) d2 Q(t) onds. Once we have our methodology in
+L = 0, (7.1) place, you will see the wonders we can per-
C dt2
form with AC circuits. Let’s get started by
where, in the second line, we’ve used the fact organizing what we know already.
that (with our choice of signs), I = ≠dQ/dt. We can take our AC EMF source and at-
Rearranging, we recognize this as the differ- tach it to the components we already know,

67
 LC circuits Princeton University Physics 104 Spring 2024
xample 4. Let us consider a series RLC circuit such as the one shown (note that
r a series arrangement the order of the parts around the loop doesn’t a!ect the
for example, in the circuit shown in Fig. 7.2. You even saw this particular equation – it’s a
uations). Recall that the complex impedances of the inductor, the resistor and the
Applying
pacitor are given by ZL =the
iωL,Loop
ZR =Rule,
R andtaking
ZC =positive
1 cur- damped, driven harmonic oscillator. We will
, respectively.
iωC
not assume that you recall all this, but you
L R should start by looking back over that mate-
rial, which was in Ch. 10 of the PHY103 little
textbook and PHY103 Learning Guide 10.
E = E0 cos(ωt)
C To deal with Eq. 7.5, we replace it with
a different equation that’s easier to solve, a
complex version:
ep 1. (Computation of total impedance). The circuit we are considering is obtained
Figurethese
y connecting in series 7.2: three
An elements,
AC RLC so its totalThe
circuit. new is given
impedance d2 QÂ dQÂ 1 Â
L by + R + Q(t) = E0 eiÊt . (7.6)
symbol on the left represents a source ✓ of ◆ dt 2 dt C
sinusoidally-varying EMF. 1 1
= R + i ωL − Here Q(t)
Z=Z L + ZR + ZC = iωL + R +
iωC ωC
 is a new, complex function, most
definitely not a solution to the original
rentequality
ote that in the last to be CW, defining
we used Q=to−i.
that 1/i beThe
the imaginary
charge part
Eq. 7.5. (L,our
of Z (in R, C, and E0 are the original,
se equal to ωL − on the
1
) top
is plate
called of the cap, and going around real parameters.) The reason we do this is
reactance.
the loop CW, we get
ωC
ep 2. (Polar representation of Z). Let us now compute magnitudesimple: theofreal
and phase Z. part of Q(t)
 is a solution to
We have: s the original real equation.
p ≠ L dI ≠ I(t)R ≠ 1 Q(t),
✓ ◆2
E(t) (7.4)
1
|Z| = Re (Z) dt 2 2 2
+ Im (Z) = CR + ωL −
ωC We will look for a steady-state solution to
Eq. 7.6, that is the behavior in the circuit
where we have noted that the charge Q on
nd when initial messiness has died away and we
the cap and the current are−going to be
1
Im (Z) I ωL
tan φ = = ωC
are left with a simple oscillating response.
time-dependent and (Z) = E0 R
Re E(t) cos(Êt). Rec-
As promised, this is easy. We try a com-
ognizing
s that,✓with our ◆
choice of signs, I =
that plex function Q(t)
 =Q  eiÊt , with the same
0
dQ/dt, 2
Z= R2 + ωL −
1 i tan 1 ((ωL−1/ωC)/R)
e frequency Ê as the driving EMF and Q Â a
0

dQ 2
dQ 1
ωC complex constant:
L 2
+R + Q(t) = E0 cos(Êt). (7.5)
dt total current).
ep 3. (Computing the dt C The total physical current in theQ(t) =Q
Âcircuit Â
is0 eiÊt ∆
! 2 Â
I’ll
⇣ remind
⌘ you that
e the goal E0 of “solving” such dQÂ
 eiÊt ; d Q = ≠Ê 2 Q
= Rea differential
E(t)
I(t) = Re equation=is toRe
E0 = iÊ Q 0
 eiÊt (7.7)
0
e find
(e the function
i(ωt−φ)
I(t) )= cos(ωt −
dtφ) = dt 2
Z |Z| |Z|
of time Q(t) that, when plugged into the and plug it in to Eq. 7.6:
E0
= qthree places it appears
cos(ωt −inφ)Eq. 7.5, gives an
equation
2 that
R + ωL − ωC is
1 2
true for all times t. This is a ≠Ê 2 LQ  eiÊt +iÊRQ  eiÊt + 1 Q
 eiÊt = E eiÊt .
0 0 0 0
daunting task , but in PHY103 you learned
1 C
(7.8)
how to solve it using
here φ = tan−1 ((ωL−1/ωC)/R)). complex
Note that exponentials.
the middle steps are completely general,
The choice of our form of Q(t), with the same
Â
E0
we can always go directly to I(t) = cos(ωt − φ).
1 |Z|
In fact, given the solution, it is hard to just plug Ê carried by the EMF, allowed the cancel-
it in and show it works! lation of all the time-dependent terms – a
8
Copyright 2024 Peter D. Meyers 68
Princeton University Physics 104 Spring 2024

sign that our guess was right. The resulting whereÔ both a and b are real numbers, and
purely algebraic equation means our guessed i = ≠1. Here a is called the real part of z,
solution to Eq. 7.6 works if denoted by a = Re (z), and b the imaginary
part of z, b = Im (z). A complex number
 =
Q
E0
. (7.9) is thus specified by two real numbers, a and
0
(1/C) + iÊR ≠ Ê 2 L b, and therefore it is convenient to think of
it as a two-dimensional vector, plotting the
To complete this, the real solution to our real part on the x-axis, and the imaginary
original Eq. 7.5 is Q(t) = Re (Q Â eiÊt ), and,
0 part on the y-axis. The corresponding x-y
since we are more interested in I than Q, plane is called the complex plane, shown in
I(t) = dQ(t)/dt. We note that we see the Fig. 7.3.
general result for driven oscillations: the re-
sponse (Q or I in this case) is an oscillation
at the driving frequency. We could do the al-
gebra to complete this example, but we are
not going to approach AC circuits this way. b=Im(z)
z=a+ib
Instead, we will come up with a very general
solution to (almost) all AC circuits, and then
simply use it. Since we will still use complex
arithmetic, we will start with a review of all
the complex-numbers techniques you’ll need.
φ

Note: a=Re(z)

The remainder of this chapter is derived from Figure 7.3: The complex plane
a set of notes edited by many hands2 and
used through the years in PHY 104 and 106.
One can manipulate complex numbers like
real numbers. For instance, we can add z1 =
a1 + ib1 and z2 = a2 + ib2 :
7.2 Complex numbers
review z1 + z2 = (a1 + ib1 ) + (a2 + ib2 )
= (a1 + a2 ) + i(b1 + b2 ).
Complex numbers are expressions of the form

z = a + ib, Just like with vectors, we just have to add

the components, which are here the real and
2
Including mine. imaginary parts. We can also multiply com-

Copyright 2024 Peter D. Meyers 69

Princeton University Physics 104 Spring 2024

plex numbers: Thus 1/i = ≠i.

Let’s check this result. i times its inverse
z1 z2 = (a1 + ib1 )(a2 + ib2 )
must of course be 1. We have
= (a1 a2 ≠ b1 b2 ) + i(a2 b1 + a1 b2 ),
1
i = 1 = i(≠i) = ≠i2 = ≠(≠1) = 1.
where we used i2 = ≠1. i
A useful notion is the one of the complex It works!
conjugate of z, denoted by z ú . It is obtained
by multiplying the imaginary part by (≠1),
which means we are reflecting the vector in Since a complex number can be thought of
the complex plane across the x-axis. If z = as a two-dimensional vector, it can be spec-
a + ib, then z ú = a ≠ ib. If we now calculate ified either by its components (the real and
imaginary parts), or by its length and the
zz ú = (a + ib)(a ≠ ib) = a2 + iab ≠ iab + b2 angle it makes with the x-axis (see Fig. 7.3).
We already saw that the length, also called
= a2 + b 2 ,
the magnitude
Ô and denoted by |z|, is given
we by |z| = zz . From Fig. 7.3, we see that
ú
Ô see that zz is a positive real number, and
ú

zz ú is just the length of the vector z in the the angle is given by

complex plane.
Im (z)
To calculate the quotient of two complex tan „ = .
Re (z)
numbers, we want to write the result in the
form a + ib, notably with no i’s in the de- In particular,
nominator. To do this, we multiply both the
denominator and the numerator by the com- z = |z| (cos „ + i sin „) .
plex conjugate of the denominator:
This can be conveniently rewritten, making
z1 a1 + ib1 (a1 + ib1 )(a2 ≠ ib2 ) use of Euler’s formula:
= =
z2 a2 + ib2 (a2 + ib2 )(a2 ≠ ib2 )
a1 a2 + ia2 b1 ≠ ia1 b2 + b1 b2 ei„ = cos „ + i sin „. (7.10)
=
a22 + b22
This formula can be derived by a Taylor ex-
a1 a2 + b 1 b 2 a2 b1 ≠ a1 b2
= +i 2 . pansion of both the exponential and the sine
a2 + b 2
2 2
a2 + b22
and cosine. It tells us that the complex num-
Example: We calculate 1/i. The complex ber ei„ is a vector of length 1 that makes an
conjugate of i is ≠i, hence we get angle „ with the x-axis. Hence we see that
any complex number z can be written as
1 (1)(≠i) ≠i
= = = ≠i. z = |z|ei„ ,
i (i)(≠i) 1

Copyright 2024 Peter D. Meyers 70

Princeton University Physics 104 Spring 2024

Ô2
|z| being the magnitude of z and „ being the We know that i must be equal to i.
angle between z and the x-axis. Let’s check that our calculation was correct:
A B2 A B2
1 1 1 1 i2
Example: Let’s calculate the magnitude Ô + iÔ = + 2i Ô +
2 2 2 2 2
andÔangle of z = ≠1 + i. The magnitude
1 1
is zz ú , and zz ú = (≠1 + i)(≠1 ≠ i) = = +i≠ = i.
2. The angle is determined by tan „ = 2 2
Im (z)/Re (z) = 1/(≠1) = ≠1. There are
many „’s satisfying this equation, namely 7.3 Solving AC
„ = ≠fi/4 + nfi for any integer n. Since
we know that z has negative real part and circuits using
positive imaginary part, „ must lie between complex impedance
fi/2 and fi, hence „ = 3fi/4, so
Ô Alternating currents of a given frequency can
z = ≠1 + i = 2ei3fi/4 .
be described by two numbers, their ampli-
tude (a magnitude) and their phase (an an-
This representation is very useful, for in-
gle). In this respect, they are just like com-
stance for multiplying or dividing two com-
plex numbers. It is convenient to think of
plex numbers: if z1 = |z1 |ei„1 and z2 =
an alternating current as a two-dimensional
|z2 |ei„2 , then
vector (sometimes called a phasor) that has
z1 |z1 | i(„1 ≠„2 ) a given magnitude I0 and rotates around the
z1 z2 = |z1 ||z2 |ei(„1 +„2 ) ; = e . origin at a given frequency f = Ê/(2fi). The
z2 |z2 |
observed, real current is the x-component of
In this way we can also easily take roots of this vector, which oscillates in time. The
complex numbers: same applies to an alternating voltage E(t) =
Ô Ò Ò E0 cos Êt. We can “represent” it by a complex
z= |z|ei„ = |z| ei„/2 . quantity, denoted by E(t),
Â

Â
E(t) = E0 eiÊt , (7.11)
Ô
Example: We calculate i. We first write where E0 is the magnitude of the E(t), so
i = eifi/2 , which can be seen from Euler’s for- E0 is a positive (and, in particular, a real)
mula. (Check this yourself from the formula, number. The physically observed voltage is
but also understand it graphically.) Hence the real part of E(t),
Â

Ô 1 2
1/2 E(t) = Re (E(t))
 = E0 cos(Êt), (7.12)
i = eifi/2 = eifi/4 = cos(fi/4) + i sin(fi/4)
1 1 so the same constant E0 is both the ampli-
= Ô + iÔ .
2 2 tude of the real voltage and the magnitude

Copyright 2024 Peter D. Meyers 71

 consider the following three simple circuits. Note
To keep track of the phases, it is sometimes easier to wor
determining
quantities andthe
take current as only
the real part a function
at the endof
of time come
the calculatio
law.
consider the following three simple circuits. Note that, in each
determining the current as a function of time comes as usual fr
law.
Example 1: Circuit with resistance.
Princeton University Physics 104 Spring 2024
Example 1: Circuit with resistance.

Imaginary
of its complex representation. This alternat-

Imaginary
ing voltage leads to current that has the same
I0 (In these fig
frequency, but may be out of phase with the E(t) R I0
voltage. If we write a complex current with E(t) R
E0 Real IE00’s should
Real h
the same frequency as the driving voltage, over them:
complex cons
Â
I(t) = IÂ0 eiÊt , (7.13) In this case, E(t) and I(t) are in phase with
we can include the phase difference one another. The current flows through the
Here between
we already know the answer. Kirchho!’s law gives E − R
current and voltage by having a we
complex resistor the instant the voltage is applied. In
Here Â
EI00cos(ωt),already
then know the answer. Kirchho!’s law
the complex notation, the vectors (or pha-
constant 1 Â pointE0in the
E0 cos(ωt), then sors) representingI(t)
Â
I(t) =and E(t)
E(t) = cos(ωt).
I0 = |I0 |e
  ≠i„
= I0 e .
≠i„
(7.14) same direction. With R 1 R E0
In this case, E(t) and I(t) are in I(t)
phase E0=
with E(t)
one =
another. cos(c
The
I0 is a real constant and the use of a minus Â
E(t) = E0 eiÊt and I(t)
 = eiÊt R (7.17) R
the resistor the instant the voltage is applied. R In the complex n
sign in the phase is our convention. The ob- e e
served current is the real partInof(or
this
thephasors)
complex E(t)
case,representing
We see and I(t)and
that I(t) areE(t)in point
phase with
in the one
same ano
directio
current, the resistor the instantE(t) Âthe voltage is applied.
= I(t)
 R, (7.18)E0 iωt In th
e = E0 e
E(t) iωt
e and E(t) e =
e point in the
and I(t) e
I(t) = Re (I(t))
 = Re (I0(or phasors)
ei(Êt≠„) ) representing I(t)
hence Ohm’s Law holds also for the complex R
We see that quantities.
= I0 Re (ei(Êt≠„) ) = I0 cos(Êt ≠ „).
(7.15) e = I(t)e iωt E
e
E(t)E(t)= E
Example 2: Circuit with inductance.0 e R, and e
I(t) =
To keep track of the phases, it is some-
hence Ohm’s law holds also for the complex quantities.
times easier to work with the complex quan-
tities and take the real partWe only see
at thethat
end Imaginary
Imaginary

Example 2: Circuit with Einductance.

0 E
=i L 0 =i L
of the calculation. To illustrate this, consider I0 I0 e = I(t)
E(t) e(In these
R,(In
the following three simple circuits. Note E(t) E(t) L L
that, in each case, the equation determining I0 E0I0 Real E0 Real
I0 ’s
I0 ’s should
hence
the current as a function of time comes as
Ohm’s law holds also for the complex quantiti
over them:
over
5
usual from Kirchhoff’s Loop Rule. complex compco
Example 2: Circuit with inductance.
Example 1: Circuit with resistance.
The Loop Rule gives
dI E(t) ≠ L dI
dt
= 0, where
E(t) − dI
Here we already know theKirchho!’s
answer. law gives
The Loop
Kirchho!’s law gives L
E(t) −
we use Faraday’s law =L 0,
dt for the
where
= 0, we useweFaraday’s
where
dt potential gener-
use Far
Rule gives E ≠RI = 0, so generated
if E(t) generated
= E0 cos(Êt),
by anby ated
anbyinductor
inductor an inductor with
with awith aa changing
changing current.
current.
changing HenceHenc
current. we se
then Hence we see that
1 E0 dI E0 dI = dIE0 E0 5
cos(ωt),
I(t) = E(t) = cos(Êt). (7.16) = cos(Êt), = (7.19)
cos(ωt),
R R dt L dt L
dt L
and,Copyright
by2024 Peter D. Meyers
by integrating,
and, integrating, 72

E0 E0 E0 E0 − π/2).
I(t) =I(t) =sin(ωt) = cos(ωt
ωL ωL ωL ωL cos(ωt −
sin(ωt) =

We see
Wethat
see the
thatcurrent lags inlags
the current phase 90◦ behind
in phase the emf.
90◦ behind the(Not
emf
 Imaginary

Imaginary
E0
= i L E0 = i L
I0 I0 (In these figures,
(In these the the
figures,
E(t) L
E(t) L I0 I0E0 Real RealI0 ’s should havehave
I0 ’s should tildestildes
E0
over overthem: them:they they
are are
complex constants.)
Princeton University Physics 104complex constants.) Spring 2024

chho!’s law and,

Kirchho!’s gives E(t)
lawbygives − L dI
E(t) −L
integrating,
dt
=dI0,=where we use
0, where we Faraday’s
use lawchoose
= 0 If we
Q/C Faraday’s for the
law potential
positive
for the current to be
potential
dt
erated by anbyinductor
generated with
an inductor
E
a
withchanging
a
E
current.
changing clockwise,
Hence
current. we
Hence see
we
I(t) =
that
see that
dQ/dt, and we obtain
0 0
I(t) = sin(Êt) = cos(Êt ≠ fi/2).
ÊL ÊL dIE0 E
dI dQ d
0 (7.20) I(t) = = CE(t)
= =cos(ωt), cos(ωt), dt dt
We see that the currentdt
lags in
dtLphaseL90¶ be- = ≠CE0 Ê sin(Êt) = CE0 Ê cos(Êt + fi/2).
hind the EMF. (Note that the EMF is at its (7.22)
, and,
by integrating,
by integrating,
maximum at t = 0, while the current reaches
its maximum at Êt = fi/2. Plot E(t) and I(t) That is, the current leads by 90¶ (again, plot
E0 E0 E0 E0 them to see this), which is to say that the
vs. Êt to I(t)
see this.)
I(t)The
= reason
= sin(ωt) is=that,=
sin(ωt) ifcos(ωt
you cos(ωt
− π/2).
− π/2). (2) (2)
suddenly apply an external
ωL ωL voltage, the ωL
ωL cur- current must flow into a capacitor first before
rent takes a while to build up, ◦ because
◦ the it can charge up a voltage different from zero,
see
Wethat theinduced
see that current lags in phase 90 behind
EMF opposes the buildup according the
the current lags in phase 90 behindthe emf. (Note
and emf.
hence that
(Note
in the emf
that thethe
a capacitor, is voltage
emfatisitsat lags
its
ximum
maximumat t = to0,
at while
Lenz’s
t= the
0,law. So,current
while in an
the reaches
inductor
current its volt-
the
reaches maximumbehind
its maximumat the
ωtat=current.
ωtπ/2.
= π/2. E(t)E(t)
PlotPlot and and
vs. seeage
I(t)ωtvs.toωt leads
tothis.)
see the The
The
this.) current.
reason is that,
reason if you
is that, if suddenly apply
you suddenly
In the an external
apply
complex voltage,
an external
notation, this voltage,
means that,
current
the current If
a we
takestakes convert
while to this
a while tosolution
build to the
up, because
build up, complex theif E(t)
the induced
because  =emf
induced , then
opposes
emf
E0 eiÊt the buildup
opposes the buildup
ording
according notation,
to Lenz’s
to Lenz’s Â
So,= in
law.E(t)
law. ESo, iÊt
0 eaninand
inductor the voltage
an inductor leads
the voltage the current.
 leads
I(t) = CÊEthe 0e
current.
i(Êt+fi/2)
= iÊCE0 eiÊt
If weIfconvert this solution to the E0complex e e
notation, E(t) = E e and and iωt iωt
we convert this solution
E0 i(Êt≠fi/2) to iÊt complex notation, E(t) 0= E0 e
the 1
 I(t) =e = e = iÊC E(t),
 or E(t)
 = Â
I(t),
ÊL iÊL iÊC
e = e E= 1E0 Â i(ωt−π/2)
0 i(ωt−π/2) E0 iωtE0 Âiωt 1 e1 e eor E(t)
e= iωL e e (7.23)
I(t) I(t) =e E(t),e or= E(t) Â = =e iÊL= (7.21)
eI(t), = E(t),E(t),
or E(t) = I(t),
iωLI(t),
ωL ωLiÊL iωL iωL iωL iωL
where we used that eifi/2 = i. Again, this
where we used e ≠ifi/2
= ≠i = 1/i. Note that looks likejust
Ohm’s law, this
rewhere we used
we used −iπ/2 −iπ/2
ethis looks −i==
e = just −i1/i.
like = 1/i.
Ohm’s Note
Law Note
withthat
that R re-this
this looks looks
just like like
Ohm’s lawtime
Ohm’s law
withwith
R “R”=
with R
1/(iÊC).
replaced
aced byplaced
by iωL, iωL,
whichby
which
is which is imaginary!
is imaginary!
imaginary!
iÊL,
In all of these three cases, we could think
Example
ample Example
3: Circuit
3: Circuit 3:with
with Circuit with capacitance.
capacitance.
capacitance. of the alternating voltage as complex, E(t)
 =
E0 e , and the corresponding complex cur-
iÊt

E0 −i E0 −i rent is then just obtained by a general form

=
I0
=
C I0 C of Ohm’s Law:
I0 I0 (In these
(In these figures,
figures, the the
E E C C
E0 EReal
0 I0 ’s should
RealI0 ’s should have
Â
E(t)
have=Z
tildes
Â
I(t),
tildes (7.24)
over over
them: them:they they
are are
or, equivalently, after dividing by eiÊt ,
complexcomplex constants.)
constants.)
E0 = Z IÂ0 . (7.25)
Choosing
osing the topthe top plate
plate
Choosing of top
the
ofplate
the the ofcapacitor
capacitor to have
to have
the capacitor
charge
tocharge
+Q,
+Q,seen,
As we’ve
Kirchho!’s
Kirchho!’s law
givesgives
law number,
IÂ0 is a complex car-
E(t)
− − Q/C
Q/C = we
= 0have
If 0charge
Ifchoose
we+Q, choose
the positive
Loop
positive Rule current
gives
current to be
E(t) ≠toclockwise,
be clockwise,
rying the phase I(t) = dQ/dt,
I(t) difference
= dQ/dt, and and
between weE.
I and
we
obtain
ain
dQ dQd d
I(t) =I(t) = = = CE(t) CE(t)
= −CE=Copyright
−CE 2024 Peter
0 ω sin(ωt)
0 ω sin(ωt) = CE D.
=0 ω
CE Meyers
0 ω cos(ωt
cos(ωt + π/2).+ π/2). (3) (3)73
dt dtdt dt
◦ 90◦ (again, plot them to see this), which is to say that
tThat is, current
is, the the current
leadsleads
by 90by (again, plot them to see this), which is to say that
the current
current mustmust
flow flow
into into a capacitor
a capacitor first first before
before it canit charge
can charge
up aup a voltage
voltage di!erent
di!erent
m from
zero,zero, and hence
and hence the voltage
the voltage lags lags
behind behind the current.
the current.
Princeton University Physics 104 Spring 2024

Here Z is the impedance, which is a complex Since

number! (Note: in keeping with common us- IÂ0 =
E0
=
E0 ≠i„
e , (7.27)
age, we will use Z (without a tilde) for com- Z |Z|
plex impedance. We will be careful to use
|Z| for its (real) magnitude.) we see that the magnitude of IÂ is given by
I0 = |IÂ0 | = E0 /|Z|, and the phase differ-
In Example 1, Z = R, which is real, while ence between the current and the voltage is
in Examples 2 and 3 it is imaginary, namely „. The actual, physical current is
Z = iÊL and Z = 1/(iÊC) = ≠i/(ÊC), re-
spectively. We add this information to our E0 iÊt
table of circuit components in Fig. 7.4. I(t) = Re (I(t))
 = Re (IÂ0 eiÊt ) = Re ( e )
Z
(Table dragged E0 iÊt E0
Symbol Name Voltage Impedance = Re ( e )= Re (ei(Êt≠„) )
over pictures)|Z|ei„ |Z|
Update 2022:
E0 add
wire ∆" ≈ 0 =
AC EMF cos(Êt ≠ „). (7.28)
|Z|
% = 0 or
switch
∆" = 0
7.4 The general
battery ∆" = ℰ
procedure
ℰ
∆" =
AC EMF
ℰ! cos +,
Exactly the same procedure works for solving
+"
capacitor ∆" = −./0 1/(3+0) more complex circuits. For instance, if the
# elements are in series, we just have to add
resistor ∆" = −%5 5 the impedances, and apply the general form
" of Ohm’s Law. Summarizing: for almost any
7% circuit driven by an EMF E(t) = E0 cos(Êt),
inductor ∆" = −6 3+6
L 7,

ground "=0
1. Find the total impedance Z by adding
individual impedances in series and/or
parallel like resistors.
Figure 7.4: Table of circuit components, includ-
ing impedance. 2. Write Z in polar form: Z = |Z|ei„ .

In general, we can decompose Z into its 3. Then, the actual, physical current is
magnitude and its phase, i.e.,
E0
I(t) = cos(Êt ≠ „). (7.29)
Z = |Z|e . i„
(7.26) |Z|

Copyright 2024 Peter D. Meyers 74

 Princeton University Physics 104 Spring 2024

4. To solve the circuit and find the volt- Step 2. (Polar representation of Z)
age across a component, use the com- We now compute the magnitude and phase
plex versions of Kirchhoff’s laws and of of Z. We have:
Ohm’s Law (7.24), then take the real Ò
part at the end of the computation. |Z| = Re (Z)2 + Im (Z)2
Û
3 42
1
We shall illustrate this general procedure = R2 + ÊL ≠ (7.31)
ÊC
by explicitly working out a number of more
complicated examples. and
1
Example 4. Let us consider the series RLC Im (Z) ÊL ≠ ÊC
tan „ = = (7.32)
circuit of Fig. 7.2, reproduced below. Recall Re (Z) R
C circuits
that the complex impedances of the induc-
ample 4. Let tor, so shown
that (note that
the resistor
us consider and RLC
a series the capacitor are as
circuit such given
the one
a series arrangement the order of the parts around1 the loop doesn’t Û
by ZL = iÊL, ZR = R and ZC = iÊC , re- a!ect
3 the 42
and the 1
Z = R + ÊL ≠
ations). Recall that the complex impedances of the inductor, the resistor
ei tan ((ÊL≠1/ÊC)/R) ,
≠1
2
spectively. 1
acitor are given by ZL = iωL, ZR = R and ZC = iωC , respectively. ÊC
(7.33)
L R though there is rarely reason to write it out
this way.
E = E0 cos(ωt)
C Step 3. (Compute the total current)
The total physical current in the circuit is
A B
1 2 Â
E(t)
p 1. (Computation of total impedance). The circuit we are considering is obtained
Step 1.
connecting in series (Compute
these the total
three elements, impedance)
so its
I(t) = Re I(t) = Re
total impedance is given by
Â
Z
The circuit we are considering is obtained✓by ◆
E0 E0
1 1
Z =Z connecting
L + ZR + Z = iωLthese
inC series +R+ three elements,
= R + i so
ωL − = Re (ei(Êt≠„) ) = cos(Êt ≠ „)
iωC ωC |Z| |Z|
its total impedance is given by
0 E
=ofÚZ (in our
te that in the last equality we used that 1/i = −i. The imaginary part
1 1 22 cos(Êt ≠ „)
Z = ZL + ZR + ZC = iÊL + R +
1
1
e equal to ωL − ωC ) is called reactance.
R + ÊL ≠ ÊC
2
p 2. (Polar representation of3Z). Let us4now compute iÊCmagnitude and phase of Z.
have: 1 (7.34)
p= R + i ÊL ≠ s ✓ (7.30) ◆2
ÊC
|Z| = Re (Z)2 + Im (Z)2 = R2 + ωL −
1
ωC where „ = tan ((ÊL ≠ 1/ÊC)/R)). Note
≠1
Note that in the last equality we used 1/i = that the middle steps in Eq. 7.34 are com-
d ≠i.3 Im (Z) 1
ωL − ωC pletely general, so we can always go di-
tan φ = =
3
I will mention thatRe
the(Z)
imaginary R rectly to I(t) = |Z|
part of Z (in E0
cos(Êt ≠ „).
1
that our cases equal to ÊL ≠ ÊC ) is called reactance. In
this class we won’t ◆2 because we are not Step 4.
✓ use this term (Solve the circuit - Compute
1
afraid
Z =of complex
R2 + numbers. ei tan ((ωL−1/ωC)/R)voltages across the circuit elements)
1
ωL −
ωC

p 3. (Computing the total current). The totalCopyright 2024 Peter

physical current D. circuit
in the Meyersis 75
!
⇣ ⌘ Ee(t) E0 E0
I(t) e
= Re I(t) = Re = Re (ei(ωt−φ) ) = cos(ωt − φ) =
Z |Z| |Z|
E0
= q cos(ωt − φ)
1 2
R2 + ωL − ωC
Princeton University Physics 104 Spring 2024

By Ohm’s Law we have: and

VÂL (t) = ZL I(t),
 E0
VC (t) = Re ( ei(Êt≠„) )
VÂR (t) = ZR I(t),
 iÊC|Z|
E0
VÂC (t) = ZC I(t),
 (7.35) = Ú 1 22 cos(Êt ≠ „ ≠ fi/2).
1
ÊC R2 + ÊL ≠
or equivalently ÊC
(7.39)
VÂL0 = ZL IÂ0 ,
VÂR0 = ZR IÂ0 , From these equations, as well as from the
equation for the current (Step 3 above), we
VÂC0 = ZC IÂ0 , (7.36) can see that the amplitudes of the volt-
where I(t)
 = IÂ0 eiÊt , VÂL (t) = VÂL0 eiÊt , etc., ages across the circuit elements and the
and in general IÂ0 , VÂL0 , etc., are complex con- amplitude of the current depend on the
stants. The physical voltage across the in- frequency Ê. This is typically one of the
ductor is: features of a circuit we are most interested
in.
VL (t) = Re (VÂL (t)) = Re (ZL I(t))
 (7.37)
and, using ZL = iÊL and I(t) Â = E(t)/Z,
 we Limiting behavior as Ê æ 0, Œ
can rewrite this as We can often isolate important aspects of
iÊLE0 i(Êt≠„) the frequency dependence by looking at lim-
VL (t) = Re ( e )
|Z| its as Ê æ 0 and Ê æ Œ. In the RLC circuit
ÊLE0 i(Êt≠„+fi/2) of Example 4, we can see from the last line
= Re ( e ) of Eq. 7.34 that if Ê æ 0, the current goes to
|Z|
ÊLE0 zero because the 1/(ÊC) in the denominator
= cos(Êt ≠ „ + fi/2) blows up. If Ê æ Œ, the current goes to
|Z|
zero because of the ÊL in the denominator.
ÊLE0
=Ú 1 22 cos(Êt ≠ „ + fi/2),At other frequencies, the current is not zero,
R + ÊL ≠ ÊC
2 1
and we’ll revisit that in Sec. 7.5.
(7.38) It is generally not necessary to solve
with „ the phase angle of the total current the circuit to find this behavior. Knowing
found above. Similarly we get the impedance of the circuit elements (Ta-
ble 7.4), and taking to heart the notion that
VR (t) = Re (
RE0 i(Êt≠„)
e ) we can treat the elements as (complex) re-
|Z| sistors, we can extract this behavior directly
RE0 from the schematic. A resistor has no fre-
=Ú cos(Êt ≠ „)
1
1
22 quency dependence. An inductor has |ZL | æ
R2 + ÊL ≠ ÊC Œ as Ê æ Œ. (Physically, this is because the

Copyright 2024 Peter D. Meyers 76

Princeton University Physics 104 Spring 2024

current is rapidly changing, making a large quency, the capacitor behaves like a wire,
opposing EMF.) It thus acts like a break in so we expect the output VC to be zero at
the circuit. Looking at the schematic for the high frequencies. At low frequencies, the
RLC circuit, we can see that this would block impedance is large. Thus the current will
the current in the whole circuit. For small Ê, be small, the voltage drop across the resistor
|ZL | æ 0, and the inductor behaves like the will be small, and VC ¥ Vapp . So this de-
wire it is. vice blocks high frequencies and passes low
Since ZC has Ê in the denominator, a ca- frequencies.
pacitor behaves oppositely to the inductor. We can proceed with the full solution fol-
At small Ê it behaves like an open circuit. lowing the General Procedure: we first com-
(Consider the limiting case of Ê = 0, that pute the total impedance of the circuit, in
is, a DC circuit. The steady state is I = 0.) cartesian and in polar form; we then com-
The cap’s impedance goes to zero at high fre- pute the complex current I(t)
 using the com-
quencies, behaving like a wire. (Physically, plex Ohm’s Law, and finally we find the volt-
with little time to charge in one quarter cy- age across the capacitor by writing VÂC (t) =
cle, the potential across it stays near zero.) Â
ZC I(t) and taking its real part.
It is a good practice to do this exercise The total impedance is just Z = R + ZC ,
before diving into a full solution. with ZC = 1/(iÊC) = ≠i/(ÊC). The
impedance in polar form can be written as
Example 5. Let’s look at the RC circuit Û
shown in the figure. 1
Z= R2 + ei„ (7.40)
(ÊC)2
R
so that

C VC Vapp
Vapp IÂ0 = Ò e≠i„ (7.41)
1
R2 + (ÊC)2

with „ = tan≠1 (≠1/ÊRC) and I(t)

 = IÂ0 eiÊt .
We have a capacitor and a resistor in se- The complex amplitude of the potential
ries with an AC generator that applies a difference across the capacitor is
voltage of amplitude Vapp , and we are asked
for the amplitude of the potential drop, VC , Vapp e≠i„
across the capacitor, which is the output of VÂC0 = ZC IÂ0 = Ò
1
iÊC R2 + (ÊC)2
the device.
Vapp
We start by looking at the limiting fre- =Ò e≠i„≠ifi/2 (7.42)
quency behavior. As we’ve seen, at high fre- 1 + (ÊRC)2

Copyright 2024 Peter D. Meyers 77

Princeton University Physics 104 Spring 2024

The physical voltage across the capacitor is The amplitude of the current is
then
E0
Vapp I0 = Ú 1 22 (7.45)
VC (t) = Ò cos(Êt ≠ „ ≠ fi/2) R2 + ÊL ≠ 1
1 + (ÊRC)2 ÊC

(7.43) which is maximal when the denominator is

with „ = tan≠1 (≠1/ÊRC). Note that for minimal, that is, when the parenthesis inside
large frequencies (i.e., large Ê) the amplitude the square root in the denominator vanishes:
of the physical voltage across the capacitor is
very small or, more formally: 1
ÊL ≠ =0 Resonance Condition
ÊC
Vapp (7.46)
lim Ò =0 (7.44) The resonance frequency is the solution to
ÊæŒ
1 + (ÊRC)2
the resonance condition equation, that is
So, as we predicted, the output voltage of the
1
device is small at high frequencies and large Ê = Ê0 = Ô . (7.47)
at low frequencies, which is why the device LC
is called a low-pass filter.
Ê0 is the oscillation frequency of the ideal
LC circuit with no resistance analyzed in
Sec. 7.1.
7.5 Resonance
We can rewrite the amplitude of the cur-
rent in terms of Ê0 :
In the examples above we found that the am-
plitude of the physical current flowing in the ÊE0
I0 = Ò . (7.48)
circuit depends on the frequency Ê of the AC L (Ê 2 R2 /L2 ) + (Ê 2 ≠ Ê02 )
2
generator. In some cases the current has a
pronounced maximum (or sometimes a mini- The phase between current and voltage is
mum) for a special value of Ê. This is called given by (show this)
resonance. In order to understand this phe- C D
nomenon, let us look back again at the series L (Ê 2 ≠ Ê02 )
tan „ = . (7.49)
RLC of Example 4. (Note: all quantities in ÊR
this section are real.) So at resonance the amplitude of the current
As we discussed above, the total current is maximal and the output current and input
in the circuit is given by Eq. 7.34, repeated voltage are in phase. These two quantities
here: are plotted, as a function of Ê, in Fig. 7.5.
E0 Far from the resonance, the amplitude of the
I(t) = Ú 1 22 cos(Êt ≠ „). current in the RLC circuit goes to zero, and
1
R2 + ÊL ≠ ÊC the phase goes to ±fi/2.

Copyright 2024 Peter D. Meyers 78

Princeton University Physics 104 Spring 2024

and let us denote by Z = iX its impedance

(we will have X = ≠1/(ÊC) for a capaci-
tor and X = ÊL for an inductor - note that
in both cases X is real). The complex volt-
age VÂ (t) = VÂ0 eiÊt across this element will be
VÂ (t) = Z I(t).
 Using Z = iX, V (t) will be
equal to iI(t)X:
Â

VÂ (t) = iX I(t)
 = iXI0 ei(Êt≠„)
= XI0 ei(Êt≠„+fi/2) . (7.50)
We can read from this equation that the
phases of voltage and current differ by fi/2,
that is voltage and current are out of phase.
The physical current and voltage are
I(t) = I0 cos(Êt ≠ „),
Figure 7.5: Amplitude and phase of I0 near res- V (t) = XI0 cos(Êt ≠ „ + fi/2)
onance as a function of angular frequency. Note
= ≠XI0 sin(Êt ≠ „) (7.51)
the log scale for Ê/Ê0 .
so that the power dissipated by the circuit
element at time t is P (t) = V (t)I(t) =
7.6 Power ≠XI02 sin(Êt ≠ „) cos(Êt ≠ „). This means
that the average power dissipated by the cir-
In this section you will learn about power cuit element over a period T = 2fi/Ê is zero,
in AC circuits. The physics is no different because
from what we learned before, P = IV , but 1
now all are functions of time. For reasons Pav = ≠X|I0 |2
T
that will become clear soon, we will be in- ⁄ T
terested mainly in the power averaged over ◊ dt sin(Êt ≠ „) cos(Êt ≠ „)
0
one or many full cycles of the oscillation. = 0. (7.52)
The first question we want to understand
is the following: What is the power dissi- (You should poke at this until you under-
pated by an inductor or by a capacitor in stand why this expression represents the
an AC circuit? The answer to this question average power. Then show that the in-
can be understood in terms of the complex tegral is zero.4 ) The conclusion is that
notation introduced above. Let there be a 4
The trig identity sin x cos x = 12 sin(2x) is help-
(complex) current I(t)
 = IÂ0 eiÊt = I0 ei(Êt≠„) ful. What is the average of sin(2x) over one period
flowing through a capacitor or an inductor, T?

Copyright 2024 Peter D. Meyers 79

Princeton University Physics 104 Spring 2024

there is no average power dissipation impedance of the full circuit. In our nota-
through elements with purely imagi- tion, working with real quantities only,
nary impedance, like capacitors or in-
P (t) = I(t)V (t) = I0 cos(Êt ≠ „)E0 cos(Êt).
ductors. Of course, energy does indeed flow
(7.55)
into and out of such a circuit element during
A trig identity makes this
each cycle, but ideal capacitors and induc-
tors (with no resistance) store energy, they P (t) = I0 [cos(Êt) cos „ + sin(Êt) sin „]E0 cos(Êt)
don’t dissipate it. = I0 E0 [cos2 (Êt) cos „ + cos(Êt) sin(Êt) sin „].
On the other hand, since current and volt- (7.56)
age are in phase in resistive elements, the
As we have seen, when we average over a
instantaneous power dissipated by a resistor
period of the oscillation,
will be P (t) = V (t)I(t) = RI02 cos2 (Êt ≠ „),
so that [cos2 (Êt)]av = 1/2; [cos(Êt) sin(Êt)]av = 0.
(7.57)
1⁄T 1 So
Pav = RI02 dt cos2 (Êt ≠ „) = RI02 ,
T 0 2 1
(7.53) Pav = I0 E0 cos „. (7.58)
2
where we used that the average of cos (Êt ≠2

„) over a period is 1/2. (An easy way to The quantity cos „ is called the power factor.
see this is to note that cos2 ◊ + sin2 ◊ = 1 and At resonance in the series RLC circuit, „ = 0,
that cos2 and sin2 have the same average over and the power factor is 1.
a period because they are the same function Using the fact that cos „ = Re (Z)/|Z|
shifted horizontally.) (see Fig. 7.3) and the generalized Ohm’s Law
The conclusion of the previous discussion result that I0 = E0 /|Z|, we can write this
is that, on average in an AC circuit, power same result in many forms, each useful in
is dissipated only by resistive elements. different circumstances:
Thus one way to compute the total average 1 1 Re (Z)
Pav = I0 E0 cos „ = I0 E0
power dissipation in an AC circuit is to find 2 2 |Z|
the amplitude of the current IR0i in each re- 1 Re (Z) 1 2
sistor Ri and sum up = E02 = I Re (Z). (7.59)
2 |Z|2 2 0

1ÿ 2 Note that E0 and I0 are the maximum values

Pav = I Ri . (7.54) of the physical voltage and current of the full
2 i R0i
circuit, and Z refers to the full circuit (and
not just the resistors). It is a useful exercise
There is also a way to treat the power to see that this expression gives the same re-
globally, using the total current provided sult as just taking I and V in the resistor in
by the oscillating EMF source and the Example 5.

Copyright 2024 Peter D. Meyers 80

Princeton University Physics 104 Spring 2024

Because of the factor of 1/2 that is present we obtain for the power as a function of the
in all of these equations, one frequently ab- frequency Ê
sorbs that factor into the current or voltage
Ô 2
by defining an Irms or Vrms which is just 1/ 2 Pav =
Erms R
. (7.63)
times less than the I0 or V0 we have used to L2 (Ê 2 ≠ Ê0 )2 /Ê 2 + R2
2

represent the amplitude or maximum value

Note that the power is a function of Ê, at-
of our complex vectors, so
taining its maximum precisely at the point
1 1 where the denominator is minimum, so that
Irms = Ô I0 , Erms = Ô E0 . (7.60) Pav is maximal at resonance, Ê = Ê0 .
2 2

Therefore, if you were to look at Erms = 120

volts from a household circuit on your lab-
oratory oscilloscope, you would see that it
would
Ô actually reach a peak level of 120 ◊
2 = 169.7 volts at the maximum of the
cosine curve on the scope (as promised in
Sec. 7.1).
Expressing the average power in these
quantities, we obtain

Re (Z)
Pav = Irms Erms cos „ = Irms Erms
|Z|
2 Re (Z) 2
= Erms = Irms Re (Z). (7.61)
|Z|2

Example. Let us consider again the series

RLC circuit. We obtained above Z = R +
iÊL ≠ i/(ÊC), and hence
3 4
2 2 1 2
|Z| = R + ÊL ≠
ÊC
2 1 22
L
= R2 + 2 Ê 2 ≠ Ê02 , (7.62)
Ê
Ô
where Ê0 = 1/ LC is the resonance fre-
quency. Inserting this expression into (7.61)

Copyright 2024 Peter D. Meyers 81

Chapter 8: Maxwell’s Equations and
Electromagnetic Waves
8.1 A problem with
!
Ampère’s Law
I I I
About 160 years ago, as the United States
fought to throw off the barbarity of slavery, C
S a b
James Maxwell, a Scottish theoretical physi- C
cist, set out to take all that was known about
electricity and magnetism and put it into a Figure 8.1: Ampère’s Law applied to a long,
rigorous, internally-consistent, mathematical straight wire carrying a current I.
framework. What was known was roughly
what we have studied so far. a) b)
x y x y
What Maxwell found was an inconsis-
I I I I
tency, rooted in Ampère’s Law. It seemed !
like a minor detail, but Maxwell fixed it, C, S Cʹ, Sʹ
making what is likely the outstanding the-
oretical discovery between Newton and Ein-
stein. To see the problem, we start with a Figure 8.2: An unsteady current.
review of Ampère’s Law, taking the simplest
case, a long, straight wire, initially treated in
Sec. 5.4.1. We apply Ampère’s Law, Maxwell’s
C, S discrepancy manifests itself in
Cʹ, Sʹ
j a number of ways.
Cʹ All involve a current that
C Cʹ
B
˛ · d˛¸ = µ0 Ithru , (8.1) is not “steady”. A steady current is time-
C
independent and has no charge building up
as shown in Fig. 8.1. The left-hand side anywhere. From the discussion surrounding
(LHS) of the equation is a line integral Fig. 4.2, we note that, for a circuit without
around a closed path C, which we choose to branches, steady current means that I is the
be a circle centered on the wire. Ithru on the same everywhere. Figure 8.2a) shows a cur-
RHS is the current going through the sur- rent that is not steady: the wire has a small
face S spanning C. Symmetry and our gen- gap between the two black dots. The setup
eral understanding of B˛ allow us to remove B
˛ shown in the figure is assumed to be part
from the dot product and integral and solve of a circuit, and the current is I everywhere
for it in terms of the current I and the dis- in the circuit except in the gap, where it is
tance from the wire. clearly zero.

82
 Princeton University Physics 104 Spring 2024
a b
So what’s the problem? Consider the Fig. 8.2b). There we were comparing calcu-
magnetic fields at the points marked x and lations of B
˛ at two different places. Here we
y. If the gap is small, they can’t be that dif- are comparing two calculations of B ˛ at the
ferent. (It wouldn’t be that hard to calculate same place, the point x. And yet, the an-
them exactly using Biot-Savart.) Now imag- swers are different due to the different Ithru ’s
ine a calculation via Ampère’s Law using the for S and S Õ , I and 0, respectively.
two Ampèrian loops shown in Fig. 8.2b). The
situation is no longer symmetric enough to x
extract B˛ from the dot product and inte-
gral, but that isn’t the issue. We already I I
know that the two magnetic fields are simi-
lar. Ampère’s Law should still be true, and C
the problem is with the RHS, specifically,
Sʹ
S
Ithru . For the loop on the left, Ithru = I,
while for the loop on the right, Ithru = 0 – Figure 8.4: The two surfaces, S (purple) and S Õ
not similar at all! (aqua), deployed in our non-steady current.

! ! Maxwell did not just find this problem,

he also proposed a solution. He noted, as
I I I you may have, that with the current I ev-
C C
C, S erywhere except in the gap, the two black
S S Cʹ, Sʹ
dots denoting the ends of the wires must be
Cʹ charging up, the left oneCʹpositive and the
Sʹ
right one negative. Perhaps inspired by Fara-
Figure 8.3: A long, straight wire carrying a cur- day’s Law, Maxwell suggested that we aug-
rent I. The two surfaces, S (purple) and S Õ ment the RHS of Ampère’s Law:
(aqua), span the same curve C.
j A B
d„e
! poke at this weak spot a bit more.
We can B
˛ · d˛¸ = µ0 Ithru + ‘0 . (8.2)
C dt
We have said that the surface S that9 Ithru
goes through can be any surface spanning C, This turns out to be the correct fix, and
and this is true for steady currents. This is Eq. 8.2 is sometimes referred to as the
shown in Fig. 8.3, where Ithru = I for any Maxwell-Ampère Law. The second term in
surface spanning C. (Two are shown: the the parentheses must (and does) have units
flat S, and S Õ with a cylindrical side and a of current. It is referred to as the displace-
single capped end). ment current, Id © ‘0 d„e /dt, though the
We try the same trick with our non-steady term “displacement” here has no meaning
current in Fig. 8.4. Note the distinction from anymore, and it isn’t a real current. In terms

Copyright 2024 Peter D. Meyers 83

Princeton University Physics 104 Spring 2024

of fields, we can write this as !

j ⁄
˛ · d˛¸ = µ0 Ithru + µ0 ‘0 d
B E
˛ · dA.
˛
C dt S #
I !" I
(8.3)
As an example, we take a look at a A
parallel-plate capacitor, making our usual !!
large-area, small-separation approximations, "
that is, that the field is uniform inside and
zero outside. The situation is sketched in Figure 8.5: Using the Maxwell-Ampère Law to
find B
˛ in a capacitor.
Fig. 8.5, where we have greatly exaggerated
separation of the plates so we can see in-
side. The plates have radius R and area A. C S
As with Fig. 8.2, we know there must be a get
magnetic field in the gap. We will find B ˛ in
the gap at r1 < R and r2 > R by applying
Eq. 8.2 to the two dashed blue circles. Az- µ0 ‘ 0 d ⁄ ˛
imuthal symmetry around the wire tells us B(r1 ) = ˛ Õ = µ0 ‘0 dE fir12
E · dA
2fir1 dt S 2fir1 dt
that B˛ is tangent to the blue circles and de-
µ0 ‘0 r1 dQ µ0 I r 1
pends only on r, so the LHS of Eq. 8.3 gives = = . (8.5)
2‘0 A dt 2fi R2
2firB(r) for r = r1 , r2 . On the RHS, there
is no current in the gap at any radius, so
Ithru = 0. We have found many times that
E = ‡/‘0 = Q/(‘0 A) inside the capacitor Comparing to Eq. 5.18, we see that this re-
and (in our approximation) E = 0 outside. sult is the same as for a fat wire with current
Noting that this means E ˛ = 0 outside the spread across it uniformly. That is, as if the
dashed black circle in Fig. 8.5, we get for displacement “current” were a real current.
r2 > R
µ0 ‘ 0 d ⁄ ˛ ˛ Õ = µ0 ‘0 dE A
B(r2 ) = E · dA
2fir2 dt S 2fir2 dt
µ0 ‘0 A dQ µ0 I
= = . (8.4)
2fir2 ‘0 A dt 2fir2 8.2 Maxwell’s Equations
This is exactly what a continuous wire would
give, an outcome of the approximations we
made for the electric field.1 For r1 < R, we When Maxwell added that term to Ampère’s
1
It should be close to what a continuous wire Law, he had the internally-consistent set of
would give, not exactly. equations he was seeking. The set of equa-

Copyright 2024 Peter D. Meyers 84

Princeton University Physics 104 Spring 2024

tions is rightly called “Maxwell’s Equations”:

j
E
˛ · dA
˛ =0 (8.8)
j jS
E ˛ = Qin
˛ · dA B
˛ · dA
˛ =0 (8.9)
j
S ‘0 S
j
d ⁄ ˛
B
˛ · dA
˛ =0 E
˛ · d˛¸ = ≠ B · dA
˛ (8.10)
S C dt S
j ⁄ j
˛ · d˛¸ = ≠ d d ⁄ ˛
E B
˛ · dA˛ B · d¸ = µ0 ‘0
˛ ˛ E · dA.
˛ (8.11)
C dt S C dt S
j
d ⁄ ˛
B · d¸ = µ0 Ithru + µ0 ‘0
˛ ˛ E · dA.
˛ The µ0 ‘0 comes about from our choice of
C dt S
(8.6) units. The minus sign in the third equation,
When you add to these four equations the Lenz’s Law, we know is necessary to conserve
Lorentz Force Law energy. Otherwise the E ˛ Ωæ B ˛ symmetry
is complete (and lovely).

F
˛ = qE
˛ + q˛v ◊ B (8.7)
8.3 From Maxwell to
˛

electromagnetic
for the force on a charge q, you have the com- waves
plete theory of electricity and magnetism:
everything else can be derived from these.
There is another feature to note in Maxwell’s
Note the parallel structure between the Equations, embodied in Eqs. 8.10 and 8.11.
equations for E˛ and B.
˛ The departure from As we saw in Ch. 6, Eq. 8.10, Faraday’s
this parallel form comes from the “source Law, says that a changing magnetic field
terms,” the explicit reference to only electric makes an electric field. Now we see in
charges and currents. Physicists have ago- Eq. 8.11, the Maxwell-Ampère Law3 , that
nized for centuries over the lack of a mag- a changing electric field makes a mag-
netic counterpart to electric charge2 , but we netic field. This suggests (and suggested to
can restore the symmetry between E ˛ and B ˛ Maxwell) that, if a changing E˛ makes B
˛ and
by considering the situation in the absence a changing B
˛ makes E,˛ maybe they can make
of electric charge and currents: Maxwell’s each other, and the whole thing can run by
Equations “in vacuum”: itself.
This is in fact the case. Here we will use
an extended example that shows in detail
2 3
The unicorn of physics: a magnetic monopole. Without the Ampère part!

Copyright 2024 Peter D. Meyers 85

Princeton University Physics 104 Spring 2024

how it comes about.4 We start by adapt- x, that is, they have no y or z dependence.
ing Eqs. 8.10 and 8.11 to the specific case at Hence
hand, where only the fields will be changing.
We can then rewrite the two equations as E
˛ = Ey (x, t) ĵ; B
˛ = Bz (x, t) k̂. (8.14)
j ⁄
ˆB
˛ y
E
˛ · d˛¸ = ≠ · dA,
˛ (8.12)
C S ˆt
Ey
j ⁄
˛ · d˛¸ = µ0 ‘0 ˆ E · dA,
˛
B (8.13)
˛ ∆x
C S ˆt 2
3 1 ∆y
where we use the partial derivative notation 4
x
to remind us that, while the fields will also O x x + ∆x
vary with position, here we are taking the
time derivative with position fixed. z
Figure 8.7: Applying Faraday’s Law (LHS).
y

We will now apply Eq. 8.12 (Faraday)

using the small rectangular path shown in
Ey Fig. 8.7. (We will eventually take the limit
x, y æ 0.) Evaluating the LHS (left-
hand side) gives
x
O j
E
˛ · d˛¸ =
C
z Ey (x + x, t) y + 0 ≠ Ey (x, t) y + 0
Bz
= [Ey (x + x, t) ≠ Ey (x, t)] y
Figure 8.6: A region of space with E ˛ in the y
direction and a B
˛ in the z direction. ˆEy (x, t)
= x y, (8.15)
ˆx
We choose for our example the setup where, in the last step, we’ve used the defi-
shown in Fig. 8.6. We have an electric field nition of a derivative,
that points in the y direction and a magnetic
field that points in the z direction. Both df f (x + x) ≠ f (x)
E
˛ and B ˛ vary in time. The fields are also © lim ,
dx xæ0 x
non-uniform spatially, but they vary only in
though we have yet to take the limit.
4
This example is an edited version of notes (in-
cluding the original figures) that accompanied the We evaluate the RHS of Eq. 8.12 using
Learning Guides. Fig. 8.8, where we have assumed that ˆ B/ˆt
˛

Copyright 2024 Peter D. Meyers 86

Princeton University Physics 104 Spring 2024
y y

dEy /dt
∆x

∆y
x 2 x + ∆x
x x
O dB/dt O ∆z 3
1
z 4
Figure 8.8: Applying Faraday’s Law (RHS). z Bz

Figure 8.9: Applying Maxwell-Ampère.

points in the +z direction.5 By going around

the path in Fig. 8.7 counter-clockwise, the The LHS gives
RHR tells us that the positive direction for j
dA
˛ is also +z, so B
˛ · d˛¸ =
C
⁄ - ≠ Bz (x + x, t) z + 0 + Bz (x, t) z + 0
ˆB
˛ -
˛ ƒ ≠ ˆBz --
≠ · dA x y, = ≠ [Bz (x + x, t) ≠ Bz (x, t)] z
S ˆt ˆt -x+( x/2)
(8.16) ˆBz (x, t)
=≠ x z. (8.19)
where, since B varies with x, we have evalu-
˛ ˆx
ated the time derivative at the center of the
patch. (The spatial variation of the time The RHS of Eq. 8.13 gives
derivative won’t matter when we take the ⁄ -
limit x, y æ 0.) ˆE
˛ -
˛ ƒ µ0 ‘0 ˆEy --
µ0 ‘ 0 · dA x z.
ˆt ˆt -x+(
We combine Eqs. 8.15 and 8.16, that is,
S x/2)

the two sides of Maxwell’s Eq. 8.12, to get (8.20)

Combining Eqs. 8.19 and 8.20,
ˆEy (x, t) ˆBz (x + x/2, t)
x yƒ≠ x y ˆBz (x, t) ˆEy (x, t)
ˆx ˆt ≠ = µ0 ‘ 0 . (8.21)
(8.17) ˆx ˆt
or, in the limit x, y æ 0,
What we have done in this section so far
ˆEy (x, t) ˆBz (x, t) is apply the two of Maxwell’s that relate E ˛
=≠ . (8.18) and B,
˛ that is, Eqs. 8.10 and 8.11, to the case
ˆx ˆt
described in Eq. 8.14, resulting in Eqs. 8.18
We repeat this reasoning with Eq. 8.13, and 8.21. We now take the derivative of
working in the x-z plane as shown in Fig. 8.9. Eq. 8.18 with respect to x and the derivative
of Eq. 8.21 with respect to t. Side-stepping
5
This could be B
˛ in the +z direction and increas- any excessive hand-wringing over exchanging
ing, or B
˛ in the ≠z direction and decreasing. the order of differentiation, we have

Copyright 2024 Peter D. Meyers 87

Princeton University Physics 104 Spring 2024

in our electromagnetic wave equation

Eq. 8.23. What speed v do you get? I get
ˆ 2 Ey (x, t) ˆ 2 Bz (x, t)
=≠ , 1
ˆx2 ˆxˆt v=Ô . (8.27)
ˆ 2 Bz (x, t) ˆ 2 Ey (x, t) µ0 ‘ 0
≠ = µ0 ‘ 0 . (8.22)
ˆxˆt ˆt2
Putting in numbers, µ0 = 4fi ◊ 10≠7 N/A2 ,
Since the terms with B
˛ are equal, we com- ‘0 = 8.854 ◊ 10≠12 F/m,
bine the two equations: 1
v=Ô = 2.9979 ◊ 108 m/s. (8.28)
µ0 ‘ 0
ˆ 2 Ey (x, t) ˆ 2 Ey (x, t)
= µ 0 ‘ 0 , (8.23)
ˆx2 ˆt2 You might recognize this as the speed of light
– Maxwell did.
and with a flourish declare, “Behold: the
Maxwell realized that electromagnetic
Wave Equation!”
waves travel at the speed of light because
This is the same equation you found in they are light!
PHY103 for transverse waves on a stretched
string

ˆ 2 y(x, t) µ ˆ 2 y(x, t)
8.4 Properties of EM
ˆx2
=
· ˆt2
, (8.24) waves in vacuum
with µ the linear mass density of the string If we had differentiated Eq. 8.18 by t and
and · the tension in the string. You learned Eq. 8.20 by x instead of the other way
then that the equation has solutions of the around, we would have gotten
form
y(x, t) = f (x û vt). (8.25) ˆ 2 Bz (x, t) ˆ 2 Bz (x, t)
= µ ‘
0 0 , (8.29)
Try ˆx2 ˆt2
Ò it in Eq. 8.24. It is a solution if v =
· /µ. You also learned how to interpret the so B
˛ and E
˛ obey the same wave equation.
fact that y is a function of two variables, x
It is convenient to consider the specific
and t, but f is a function of only the combi-
case of sinusoidal waves, presented here in
nation x û vt: f is a fixed shape that travels
three useful forms:
in the ±x direction at speed v. It is a trav-
eling wave. Ey (x, t) = E0 sin(k(x û ct))
Now try the solution = E0 sin(kx û Êt)
x
= E0 sin(2fi( û f t)), (8.30)
Ey (x, t) = f (x û vt) (8.26) ⁄

Copyright 2024 Peter D. Meyers 88

Princeton University Physics 104 Spring 2024

with c the speed of light. In our example we where we have used µ0 ‘0 = 1/c2 and, for a
chose to have E ˛ with only a y component. sinusoidal wave, Ê/k = f ⁄ = v = c.
We see that the traveling wave maintains this We can represent this electromagnetic
direction as it propagates in x. The direction wave as shown in Fig. 8.10. For clarity, in
of E
˛ in such a wave is called its polarization.
the top picture only the fields along the x
We can apply Eq. 8.21 to the wave de- axis are plotted. However, this representa-
scribed in Eq. 8.30, choosing the wave to be tion does not tell the full story. Our E ˛ and
traveling in the +x direction: B
˛ vary with x by design. However, they also
ˆBz (x, t) ˆEy (x, t) explicitly do not vary with y or z. So, for any
= ≠µ0 ‘0 point on the x axis, any other point with the
ˆx ˆt
ˆ same x, that is, any point in a plane par-
= ≠µ0 ‘0 E0 sin(kx ≠ Êt) allel to the y-z plane with the same x, has
ˆt
= ≠µ0 ‘0 E0 (≠Ê) cos(kx ≠ Êt). the same E ˛ and B˛ at any given time. Such a
(8.31) wave is called a plane wave. The bottom pic-
ture is an attempt to show this by plotting
The LHS suggests we can get Bz by inte- E
˛ and B ˛ in two such planes, located along x
grating over x. Ignoring the uninteresting at the point indicated by the cross-hairs in
constant of integration, each frame.
⁄
Bz (x, t) = µ0 ‘0 E0 Ê cos(kx ≠ Êt)dx Plane waves are an idealized form – most
Ê real sources (lightbulbs, radio antennas) ra-
= µ0 ‘0 E0 sin(kx ≠ Êt). (8.32)
k diate spherically. In that case, the radiation
We learn a lot from this! is approximately a plane wave to any viewer
a long distance from the source, as illustrated
• If E is a sinusoidal traveling wave, prop-
in Fig. 8.11.
agating in the +x direction, then B ˛ is
also a traveling wave propagating in +x.
• B
˛ in an EM wave is perpendicular to E, ˛
and we note that the wave propagates 8.5 Energy transmitted
in the direction of E˛ ◊ B.
by EM waves
˛
• B
˛ has the same frequency and wave-
length as E,˛ and, of course, the same
v = c. We have seen that the energy stored in ca-
• B
˛ is in phase with E.˛ pacitors and inductors can be considered to
• The amplitudes of E ˛ and B ˛ are related. be stored in their electric (Eq. 3.21) and mag-
netic (Eq. 6.16) fields. With EM waves, this
For this last point, from Eq. 8.32 we see that notion goes from mathematical contrivance
Ê E0 to clearly real. The warmth of the sun on
B0 = µ0 ‘0 E0 = , (8.33) your face is energy carried through 93 million
k c

Copyright 2024 Peter D. Meyers 89

Princeton University Physics 104 Spring 2024

cB

y
!
E

"
cB

"
!
x

Figure 8.10: An electromagnetic plane wave propagating in the +x direction and polarized in the
y direction. The top figure shows E˛ and B
˛ plotted only along the x axis. The bottom figure shows
that, for this wave, all points in any plane parallel to the y-z plane have the same E
˛ and B ˛ at
any time. Two such planes are shown, with the arrows labeled E and B showing the fields in each
˛ ˛
13
plane on the x axis.

Copyright 2024 Peter D. Meyers 90

Princeton University Physics 104 Spring 2024

E
y
the frame, so

(ūE + ūB )Ac t

x
I= = (ūE + ūB )c
Source z C A t D C D
‘0 2 1 2 ‘0 2 1
= E + B c= E + E2 c
2 2µ0 2 2µ0 c2
‘0
Nearly plane waves
= ‘0 E 2 c = E02 c, (8.34)
2
Figure 8.11: Far from a source, initially spheri-
cal waves become approximately plane waves.
where we used Eqs. 3.21 and 6.16, B = E/c
in a wave, µ0 c2 = 1/‘0 , and the average value
of E 2 in a sinusoidal wave is E02 /2.7 Note how
miles of vacuum by the electric and magnetic many factors of two came and went!
fields that are sunlight. As an example, the luminosity of the sun,
that is, the power emitted as EM waves, is
L§ = 4◊1026 W. This radiation goes equally
in all directions. We find the intensity of
sunlight at Earth by dividing by the area of
(! + ("
a sphere with radius equal to the Sun-earth
!=# distance:
A
L§ 4 ◊ 1026 W
I= = = 1400 W/m2 .
4fir2 4fi(1.5 ◊ 1011 )2 m2
(8.35)
Δ% = #Δ& This applies at the top of the atmosphere.
About half is absorbed or reflected away be-
Figure 8.12: Sampling energy carried by an EM fore reaching the ground, but this still leaves
plane wave. about 1 kW/m2 for use as an energy source.
Note that14 Eq. 8.35 shows the origin of the

well-known 1/distance2 dependence of inten-

We define intensity I as average power per sity, which has its root in Conservation of
area carried by a plane wave. This is illus- Energy.
trated in Fig. 8.12, where a plane wave car-
rying average6 energy density ūE + ūB passes With the intensity of sunlight, we can
a fixed frame with area A. In a time t, the use Eq. 8.34 to find the amplitudes E0 ¥
energy in the volume Ac t passes through 1000 V/m and B0 ¥ 3 µT.
A compact way to describe power trans-
6
We use a bar over a quantity to indicate an av-
7
erage. Recall this last point, discussed in Sec. 7.6

Copyright 2024 Peter D. Meyers 91

Princeton University Physics 104 Spring 2024

mitted by E
˛ and B
˛ is the Poynting vector: 8.6 Appendix
˛S © E ◊ B ;
˛ ˛
S̄ = I. (8.36) [This section is included for the curious and
µ0
is not a required part of our course. Enjoy!]
Note that ˛S poynts8 in the direction the en- ℓ/2
ergy is being transmitted. Also, using the
flashlamp black pad
the fact that E
˛ ‹B˛ in an EM wave and the
information following Eq. 8.34, we have
EB E2
S= = ;
µ0 µ0 c !#$
E2 E2 ‘0
S̄ = = 0 = E02 c = I, (8.37)
µ0 c 2µ0 c 2 Figure 8.13: A frictionless boxcar.
with I the intensity found in Eq. 8.34.
Here is an interesting consequence of light
We end the chapter by stating without
carrying momentum = energy/c. Consider a
proof9 that EM waves carry momentum as
frictionless railroad boxcar of mass (includ-
well as energy, and that the momentum car-
ing contents) M , initially at rest. A flash-
ried is the energy carried divided by c. This 15
lamp in the boxcar emits a very short flash
means that (energy per area per time)/c,
of light with energy E. Its momentum is
that is, I/c, is momentum carried per area
then E/c. After the flash is emitted, the car
per time. Consider the case where the EM
must have equal and opposite momentum,
wave is incident on a surface that absorbs it.
and thus
Then I/c is dp/dt per area, which is force per
area, which is pressure – radiation pressure, pcar E
vcar = =≠ . (8.39)
abs
Prad = I/c, (8.38) M cM
When the light hits the black pad at the
where I used the superscript abs to indicate
other end of the car, the pad absorbs the
the case where the radiation is totally ab-
light, and its energy and momentum. Hence
sorbed. Ask yourself: how is the radiation
the final velocity of the car is zero. The time
pressure modified if the wave is reflected by
it takes the flash to cross the car is, to a very
180¶ ?10
good approximation, t = ¸/c, hence the car
This is generally a small effect. The ra- has moved a distance
diation pressure of sunlight on you is about
E ¸
1 kW/m2 /c ¥ 3 µN/m2 . x = vcar t = ≠ . (8.40)
cM c
8
Funny physics joke.
9
There is no reasonably elementary derivation. This is odd: the center of mass of a closed
10
Answer: it is twice as large. system moved. That isn’t possible. We ask

Copyright 2024 Peter D. Meyers 92

Princeton University Physics 104 Spring 2024

the equally odd question: how much mass

would have to have moved from one end of
the car to the other to keep the CM station-
ary? Moving a mass m from the left end of
the car to the right end would move the CM
by xCM = m¸/M (verify this for yourself).
We thus want

xCM = ≠ x
m¸ E ¸
=
M
⇢⇢ c⇢
M⇢c
or
E = mc2 .

So there’s that.

Copyright 2024 Peter D. Meyers 93

Chapter 9: Phenomenology of Light as an
EM Wave
9.1 The electromagnetic observed in, for example, x rays and radio
waves.
spectrum

Electromagnetic waves of very different fre-

quencies are all handled nicely by the ap-
9.2 Polarization
proach of Chapter 8.1 However, in practice,
they are generated and detected using very The first thing to note about polarization is
different methods and instruments, and they that most of the light around us, from the
are described with different names, indicated sun, say, or from lamps, is not polarized in
in Fig. 9.1. In this chapter, we will focus on the sense that the wave shown in Fig. 8.10
visible light, but similar phenomena can be is. This is because most light is generated by
random processes, the result of emission from
1
Quantum mechanics also enters the picture at many independent atoms. We call such light
very short wavelengths. unpolarized, but an incoherent (random in

Figure 9.1: The electromagnetic spectrum, identifying with names various regions of frequency and
wavelength. UV and IR are Ultraviolet and Infrared. (Wikimedia commons)

94
Princeton University Physics 104 Spring 2024

phase) superposition of many polarizations Fig. 9.2c), the second polarizer is at an angle
is a better picture. We can turn this into ◊ to the first. The light comes out polarized
polarized light by passing it through a po- along the new axis with an amplitude that is
larizer, a material that allows only a single cos ◊ times the amplitude of the light before
polarization direction to pass through. We the second polarizer: this is just the compo-
will call this direction, perpendicular to the nent of the incident E˛ parallel to the axis of
direction the light is traveling, the axis of the the second polarizer. Since intensity of light
polarizer. This action is best thought of as is proportional to amplitude-squared, we can
absorbing the vector component of E ˛ that is recast this simple behavior as a general rule
perpendicular to the axis. The light coming for the intensity of initially polarized light
out is then automatically polarized parallel after a subsequent polarizer:
to the axis, as shown in Fig. 9.2.
I = I0 cos2 ◊, (9.1)
where I0 is the intensity of the light before
a
the polarizer and ◊ is the angle between the
polarization of the incident light and the axis
of the polarizer.
We can then ask: what happens to inten-
b sity of the initially unpolarized light at the
first polarizer in Fig. 9.2? Our picture of
unpolarized
O light as having all polarizations
randomly mixed suggests that the polarizer
c will yield the average of cos2 ◊, that is the
intensity of initially unpolarized light
after a polarizer is half the incident in-
Figure 9.2: Initially unpolarized light at left
tensity.
passes through various polarizers. Polarization
is indicated by the red lines, polarizer axes indi-
cated by the solid black lines.
9.3 Reflection
Once the light is polarized, we can apply
our polarizer rules to the cases in Fig. 9.2. Light incident on a perfect conductor does
In Fig. 9.2a), a subsequent polarizer with an not enter the conductor. The electric and
axis parallel to the first polarizer does noth- magnetic fields of the light apply oscillating
ing: there is no perpendicular component of forces to the free electrons at the surface of
E
˛ to absorb. In Fig. 9.2b), the second po- the conductor, which themselves emit EM
larizer is perpendicular to the polarization of waves. These waves destructively interfere
the light that hits it, and all is absorbed. In with the incident wave inside the conductor

Copyright 2024 Peter D. Meyers 95

Princeton University Physics 104 Spring 2024

(leaving E ˛ = 0 in the bulk of the conductor), from the free charges on the plates, due to
but also produce a reflected wave. This re- the “polarization”2 of the dielectric: the dis-
flected wave leaves the surface at the same placement of the charges in the atoms of the
angle that the arriving wave had, but on the material due to the electric field. This result
opposite side of the normal to the surface, is quite general. If we place a point charge
as shown in Fig. 9.3. We say, “The angle Q in a dielectric medium, the electric field
of reflection equals the angle of incidence,” around it is
◊1Õ = ◊1 .

!! !!" 1 Q
E(˛
˛ r) = r̂, (9.2)
4fiŸ‘0 r2

with Ÿ > 1 the dielectric constant defined in

Sec. 3.6.3. There is no new mystery here:
Figure 9.3: An incident plane wave (red) reflect- Q attracts opposite charges in the medium
ing from a conductor, giving an outgoing plane (still bound in their atoms), like the Q on the
wave (blue). The red and blue arrows indicate capacitor plate did, surrounding itself with a
propagation directions. The red and blue lines bit of the opposite charge.
perpendicular to the arrows indicate the “crests”
of the waves – surfaces of maximum E. ˛ The impact of this phenomenon on elec-
tromagnetic waves is strikingly simple. Go-
ing from Maxwell’s equations in vacuum,
Eqs. 8.8-8.11, to in a dielectric with a uni-
9.4 Light in transparent form Ÿ, the only change ends up being the
same replacement of ‘0 with Ÿ‘0 that oc-
media curred in Eq. 9.2.3

Common experience shows that some mate- Maxwell’s equations in a uniform dielec-
rials are transparent, that is, light can prop-
agate through them. Most such materials
are insulators: glass, some plastics, water. 2
The meaning of “polarization” here is completely
Our only discussion of such materials so far different from that in Sec. 9.2.
was in Sec. 3.6.3, where we noted that most 3
This is far from obvious. The changing E ˛ in an
real capacitors had an insulator, which we re- EM wave means that the polarization of the medium
is changing. This in turn means that the charges
ferred to as a dielectric, between their plates. bound in the atoms of the dielectric are moving in
We noted that the electric field in such a ca- an organized way: a real (time-varying) current, with
pacitor was weaker than would be calculated all its ramifications.

Copyright 2024 Peter D. Meyers 96

Princeton University Physics 104 Spring 2024

tric are then will become clear in the next section. Fig-
j ure 9.4 shows the index of refraction of a par-
E
˛ · dA
˛ =0 (9.3) ticular type of glass, measured at four wave-
jS lengths spanning the visible spectrum. The
B
˛ · dA
˛ =0 (9.4) effect is small (note the vertical scale), but
S
j
d ⁄ ˛ important.
E
˛ · d˛¸ = ≠ B · dA
˛ (9.5)
C dt S
j
d ⁄ ˛
B · d¸ = µ0 Ÿ‘0
˛ ˛ E · dA.
˛ (9.6)
C dt S
The steps to the wave equation are the same
as before – the only difference is that we re-
place ‘0 with Ÿ‘0 everywhere. The immediate
effect is that the speed of light in the dielec-
tric becomes Figure 9.4: Index of refraction of borosilicate
crown glass vs. wavelength. (Data from CRC
1 c c Handbook of Chemistry and Physics)
v=Ô =Ô = , (9.7)
µ0 Ÿ‘0 Ÿ n

where we introduce a new parameter called

Ô
the index of refraction, n = Ÿ. 4 The up-
9.5 Refraction
shot: light travels slower in a dielectric
than in vacuum, by a factor n that de- Consider a plane wave in vacuum incident
pends on the material. normally on a flat plate of glass with index
of refraction n. The wave propagating in
This is not a small effect. Glass has the glass is driven by the incident wave in
n ¥ 1.5 and water has n ¥ 1.33, easy-to- vacuum. We have seen that the response in
remember as 32 and 43 , respectively. Air, by any such driven oscillation is at the same fre-
the way, is a dielectric with n = 1.0003. This quency as the driving source. Since f ⁄ = v
is in most cases negligibly different from vac- for the waves, and the wave speeds in the
uum’s n = 1. vacuum and glass are different, we conclude
n is observed to depend on wavelength, that the wavelengths in the two media are
a property called dispersion for reasons that different: the slower wave in the glass has
a shorter wavelength than the wave in vac-
4
Historically, the effect of an insulator on light uum. In general, at an interface between two
and its effect on capacitors came from different
media with indices of refraction n1 and n2 ,
experiments at different times, so this equality is
another impressive unification of phenomena via v1 v2 c c
Maxwell’s Equations. =f = ∆ = ,
⁄1 ⁄2 n 1 ⁄1 n 2 ⁄2

Copyright 2024 Peter D. Meyers 97

Princeton University Physics 104 Spring 2024

hence In Fig. 9.5a), the incident ray and two

⁄2 n1 consecutive wavefronts (maxima or crests of
= . (9.8)
⁄1 n2 the wave) are shown in red. The two wave-
fronts are, by definition, ⁄1 apart in medium
This is true for any angle of incidence. 1. Our immediate goal is to draw these two
For non-normal incidence, that is, using wavefronts in medium 2. We marshal these
the notation of Fig. 9.3, ◊1 > 0, there is an- facts:
other effect to contend with: the difference • The point labeled i (j) is on the first
in light speeds and wavelengths in the two (second) wavefront in both media.
media results in a change in direction of the • From Eq. 9.8, ⁄2 = ⁄1 (n1 /n2 ) < ⁄1 ,
wave. We can see this via Fig. 9.5, in which a since n2 > n1 in our example.
plane wave with wavelength ⁄1 propagating
in a medium with index of refraction n1 is in- These imply that the first wavefront in
cident on a medium with index of refraction medium 2 goes through point i and is a dis-
n2 > n1 , at an incident angle ◊1 . tance ⁄2 < ⁄1 from point j along the direc-
tion the wave is traveling in medium 2. This
means that the first wavefront in medium 2
a) #! must be tangent to the dashed green circle,
!! as shown.
!!
We see that this requires that the wave
i change direction at the interface, as indicated
j ## by the green arrow in Fig. 9.5b). Our new
goal is to calculate this new direction, that is,
!# > !! ◊2 in the figure. We do this by noting the two
right-triangles that share hypotenuse h along
the interface. Each triangle has a wavelength
b) "! as one of its sides, giving
!! ⁄1 ⁄2 ⁄1 ⁄2
!! =h= ∆ = ∆
$! cos –1 cos –2 sin ◊1 sin ◊2
j $" ℎ i sin ◊2 ⁄2 n1
= = , (9.9)
!" sin ◊1 ⁄1 n2
""
!# > !! where we used Eq. 9.8 in the last step. The
last line is what we were looking for,
Figure 9.5: A plane wave (red) incident on the n1 sin ◊1 = n2 sin ◊2 , (9.10)
boundary between media of different indices of
refraction changes direction (green). known as Snell’s Law.

Copyright 2024 Peter D. Meyers 98

Princeton University Physics 104 Spring 2024

Looking at Fig. 9.5b), it is pretty easy What does happen is that all the light
to visualize this: the part of the wavefront beyond the critical incident angle ◊c , with
in medium 2 slows down, but the part in sin ◊c © n2 /n1 , is reflected, and none con-
medium 1 continues at the original speed. tinues into the second medium. This pro-
This bends the path of the light at the inter- cess yields essentially perfect reflection, bet-
face, a process called refraction. In going ter than any mirror, and is called total inter-
from a lower index to a higher one, the nal reflection. It is the basis of optical fiber
refracted beam of light is bent toward technology.
the normal.

Because the index of refraction depends

on wavelength, the angle of refraction will
also depend on wavelength. This means that
white light entering glass at an angle will
be spread into its component colors, one of
the reasons this property is called dispersion.
The data in Fig. 9.4 indicate that short-
wavelength (violet) light will be deflected
through a larger angle than long-wavelength
(red) light. Besides giving useful ways to ma-
nipulate light in the laboratory, this behavior
is also what makes a rainbow so lovely.

In entering even the most transparent of

media, some of the light is reflected. The
fractions transmitted and reflected are cal-
culable from the incident angle and indices
of refraction, but this is beyond the scope of
our course. However, there is a case that is
definitely within our scope. In going from
a higher index to a lower one (say, glass to
air), the beam of light is bent away from the
normal, ◊2 > ◊1 . This runs into mathemat-
ical “issues”. Consider a beam of light in
glass (n1 = 32 ) arriving at a boundary with
air (n2 ¥ 1) at an incident angle ◊1 . For
sin ◊1 > 23 , that is, ◊1 > 41.8¶ , Snell’s Law
gives sin ◊2 > 1, which is not happening.

Copyright 2024 Peter D. Meyers 99

Chapter 10: Interference and Diffraction of
Electromagnetic Waves
We know from our study of electric fields path labeled 1). The mirrors direct the red
what happens when multiple sources of E, ˛ path onto the second beam splitter, where
that is, charges, are present: at any given half the light is directed to the right, as is half
point in space the fields from the different the light from the blue path. (The second
charges simply add (as vectors), a property splitter directs the other half of each beam
we call “superposition.” The same is true for downward, as shown by the dashed lines la-
magnetic fields. In this chapter, we will con- beled 3 and 4.)
sider situations where the fields are those The interesting result here comes from the
in sinusoidal electromagnetic waves.1 The fact that, in the recombined outgoing beam,
physics is still just superposition. We de- the wave that took the red path has traveled
fine “interference” of two or more waves as farther than the wave taking the blue path,
“cases where superposition gives something by a distance 2s. This means that, at any
interesting.” place along the outgoing path, the phases2
of the waves we are superposing differ. Since
2fi of phase corresponds to one period of the
2
2 sinusoid or one wavelength of distance, the
" 2 phase difference between the two waves mak-
1 2
ing the outgoing beam is
ℓ 1
3 4
”
„ = 2fi , (10.1)
Figure 10.1: Beam splitters allow light to take ⁄
two different paths to get to the same place.
where ” is the difference in path length. In
Fig. 10.1, ” = 2s.
Figure 10.1 shows an example. The solid
black lines in the figure are mirrors. The To see why this matters, consider two
dashed black lines are “beam splitters,” mir- cases. In the first case, the distance s is
rors that reflect half the light and transmit 1000 times the wavelength of the light. When
the rest. At the first beam splitter (on the the beams recombine, the red-path beam has
left), the incident beam is split into the re- traveled 2000⁄ farther and the phase of the
flected beam (indicated by the red path la- 2
At any fixed location, the electric field in each
beled 2) and the transmitted beam (blue wave has the form E(t) = E0 sin(Êt + „). Êt + „
is the phase of the wave. In our applications, the
1
In this chapter, it is most convenient to consider superposed waves will have the same frequency, but
E
˛ to always be perpendicular to the page. they can have different values of „.

100
Princeton University Physics 104 Spring 2024

red-path beam has advanced by an integer (about 100 nanometers = 0.1 micron for
multiple of 2fi with respect to the blue-path blue light), the outgoing beam changed from
beam. This means that the two waves are bright to dark. This means that a device
in phase (multiples of 2fi don’t matter), and like that in Fig. 10.1 can measure changes in
their amplitudes simply add, as shown in length (in this case s) with exquisite preci-
Fig. 10.2 (top). This situation is called “con- sion. Such a device is called an “interferom-
structive interference.” eter.”4
In the second case, s increases by 14 ⁄. This
adds 12 ⁄ to the red path. The outgoing waves
are now out of phase by half a wavelength, 10.1 Thin-film
or fi radians, as shown in Fig. 10.2 (bottom).
Here there is no outgoing beam at all, a sit-
interference
uation called “destructive interference.”3
We can devise a one-dimensional setup where
a single source of sinusoidal EM plane waves
(light, in this case) can lead to interference
by using the partial reflection that always oc-
curs at the interface between two transpar-
ent media. Consider light originally in air,
incident normally on a thin sheet of glass.
As we mentioned at the end of the previous
chapter, most of the light is transmitted into
the glass, but a small amount (around 5%)
is reflected. The same thing happens at the
back surface of the glass. When this second
reflected ray reaches the front surface, most
of it is transmitted. This situation is shown
Figure 10.2: Snapshots of two identical waves, in Fig. 10.3. Note that in the region above
drawn as blue and red, propagating to the right. the glass plate, there are now two light waves
The green curve is their sum. Top: blue and traveling upward. Note also that wave 2 re-
red in phase, giving constructive interference. flected from the back surface of the glass has
Bottom: red out of phase with blue by ⁄/2, or traveled farther than wave 1, by a distance
fi radians, giving destructive interference.
” = 2s, where s is the thickness of the glass.
1
When the distance s changed by 4
⁄ 4
The LIGO experiment (the I is for “interferom-
eter”) can measure changes in a four kilometer path
3
You might ask yourself where the energy in the as small as one ten-thousandth the width of a pro-
original beams goes in this case – a good question ton, and thus observe the warping of space when a
that we will revisit. gravitational wave passes by.

Copyright 2024 Peter D. Meyers 101

Princeton University Physics 104 Spring 2024

The phase difference between the two outgo- The overall phase difference between wave
ing waves depends on how many wavelengths 1 and wave 2 is a combination of the phase
this distance is. Note that what matters here difference due to the extra path length for
is the wavelength in the glass, ⁄Õ , which is dif- wave 2, as in Eq. 10.1, and the phase changes,
ferent from the wavelength of the light in air if any, from the reflections,
⁄, as described in Sec. 9.5. Using Eq. 9.8,
with ⁄ the wavelength in air and nair ¥ 1, „ = „2 ≠ „1
⁄Õ = ⁄/nglass . = („path + „r2 ) ≠ „r1
3 4
2s
incident wave = 2fi Õ + „r2 ≠ „r1 . (10.3)
⁄
1 2
wave reflected at front wave reflected at back, In our example, „r1 = fi, since n1 = nair <
'%)! ≈ 1 transmitted at front
"!" n2 = nglass for the front reflection, and „r2 =
wave transmitted at front 2! 0, since n1 = nglass > n2 = nair for the back
"$%&' = 2%
'*+%,, ≈ 1.5 &( ! reflection. We thus have
wave reflected at back
'%)! ≈ 1 "!# 2s 2snglass
„ = 2fi ≠ fi = 2fi ≠ fi, (10.4)
⁄Õ ⁄
Figure 10.3: Thin film setup. Reflections from
the top and bottom surfaces of a thin glass plate where we have used ⁄Õ = ⁄/nglass .
interfere. How do we interpret this result? Let’s
consider an example in which the glass thick-
We must add one further twist to this: ness is one quarter the wavelength in the
reflection from an interface can itself cause glass, that is, the extra pathlength 2s =
a phase change, „r . For us, this will just be ⁄Õ /2. Applying Eq. 10.3, the extra path-
a rule to remember: for a wave going from length gives „path = fi, and we have „r1 = fi
a medium with index of refraction n1 to one and „r2 = 0, so „ = 0. This is the case
with n2 , shown in the top panel of Fig. 10.2: the two
Y reflected waves constructively interfere, and
]0, n 1 > n2 there is a lot of reflection.
„r = (10.2)
[fi, n1 < n2 ,
Now let the glass thickness be half the
wavelength in the glass. Do the calculation
that is, for light entering a medium of higher
to electromagnetism per se. A mechanical exam-
index (slower wave speed), the reflection has ple would be two strings of different mass-per-length
a phase shift of fi radians, which means it (and hence different wave speeds) knotted together.
flips over. Otherwise, there is no phase shift There is a reflection at the knot, and the reflected
from reflection.5 wave is inverted if the incident wave is going from
the lighter string to the heavier one (with a slower
5
This is a general property of waves, and not due wave speed).

Copyright 2024 Peter D. Meyers 102

Princeton University Physics 104 Spring 2024

and show that „ = fi, as in the bottom panel crests, indicated in blue), this is pretty much
of Fig. 10.2: the two reflected waves destruc- what happens. And, as expected, when the
tively interfere, and there is essentially no gap is made wider, the width of the transmit-
reflection.6 ted segment of the wave gets wider as well.
This effect is responsible for the rainbow However, when the width of the gap is of the
sheen from an oil film floating on water. The order of or smaller than a wavelength, the
varying thickness of the film means that dif- behavior is very different. As the right panel
ferent colors of light have the right wave- shows, with the gap smaller than ⁄, the out-
length to constructively interfere at different going wave spreads more broadly. In the case
places on the film.7 shown, with gap<⁄, the wave spreads in all
forward directions.8

10.2 Diffraction from a

narrow aperture
Figure 10.4 shows plane waves approaching
an opaque barrier with a gap in it. You can
imagine that you are looking down on wa-
ter waves, traveling to the right, coming to
a breakwater with a passage through it. We Figure 10.4: Plane waves (crests shown in blue),
expect that the part of the wave that hits the propagating to the right, incident on a barrier 3

wall will not go further to the right, while (yellow) with a gap. Left: gap∫⁄. Right: gap.
the section of the wave that “hits” the gap ⁄. Images captured from phet.colorado.edu.
will continue to the right. For cases like that
on the left, where the width of the gap is This spreading is called diffraction, and it
considerably greater than the wavelength of is a property of all waves.9 For now, what
the wave (the distance between consecutive we need is just the fact that for a gap. ⁄,
6 the wave spreads uniformly downstream of
The reflection from an air-glass interface is un-
desirable in lenses, and is sometimes eliminated by the gap. If this were light passing through
putting a thin “anti-reflection” coating on the lens a hole smaller than a wavelength, the inten-
with an index of refraction different from both air sity would be uniform in all directions down-
and glass. The thickness of the coating is chosen
8
so that light reflected from the coating-glass surface Try it yourself at
interferes destructively with light reflected from the https://phet.colorado.edu/sims/html/wave-
air-coating surface. The result is no reflection at all. interference/latest/wave-interference en.html .
7 9
Since the thickness varies continuously, so does When the waves represent elementary parti-
the strongly reflected wavelength, giving the rain- cles in quantum mechanics, this effect is called the
bow. Heisenberg Uncertainty Principle.

Copyright 2024 Peter D. Meyers 103

Princeton University Physics 104 Spring 2024

stream of the barrier. screen

We will be dealing primarily with EM

waves, and our typical setup will be an
#
opaque card with one or more vertical slits
in it. This is shown in Fig. 10.5. Note that $
!
the cross-section view of the single slit is es-
sentially the same setup as that of Fig. 10.4. %
Initially, we will deal with slits that are “nar- incident
row,” which will mean . ⁄ wide. plane wave
"

Figure 10.6: The two-slit setup, before any ap-

proximations.
the screen, we start with the intensity on the
screen from either slit alone. We will call the
intensity at the center of the screen from one
------- - ------- slit alone I0 .
Figure 10.5: Opaque cards with one (left) and To find the intensity with both slits, what
two (right) narrow slits. Below each card is the matters is the phase difference between the
same card, viewed in cross section (red dashed two waves reaching the screen at y, which
lines) from above. here comes entirely from the path difference.
We can identify this path difference geomet-
rically by adding the dark blue bar that
forms an isosceles triangle with the two rays
10.3 Interference pattern to the point of interest on the screen. With
this construction, the lower wave travels an
from two narrow additional distance ”, as shown.
slits Finding ” is a trig problem. You could
simply grind through it, but we will instead
Figure 10.6 shows a general view of the case simplify things by looking at the most com-
of two narrow slits separated by a distance mon case, d on the order of ⁄ and L ∫ ⁄.10
d. We will be interested in the intensity of For L ∫ d, the two rays meeting at the
the light when it reaches a screen a distance
L from the card-with-slits, say, at a point y 10
d on the order of ⁄ is where the interesting in-
from the center, as shown. terference effects are, and, for both light and ra-
dio waves, we are generally interested in intensity
To find the intensity at a given point on MANY wavelengths away from the source.

Copyright 2024 Peter D. Meyers 104

Princeton University Physics 104 Spring 2024

screen are approximately parallel; this is we end up with alternating bright and dark
shown in Fig. 10.7. With this approximation, bands, called “fringes,” on the screen.
we can find the angle ◊ in the small triangle Now, to the math. The physics is still
of the bottom figure (verify this), then the just the superposition of the fields in two EM
path difference is easy: ” = d sin ◊. waves, one from each slit. We will keep track
of the electric fields – the magnetic fields do
screen the same thing. Let’s start with the elec-
tric field at position y on the screen due
to the top slit only. Since we started with
#
a sinusoidal plane wave, this will be sinu-
$ soidally varying: E1 (y, t) = E0 sin(Êt). Be-
!
cause the wave from the lower slit has trav-
% eled farther, it has advanced further in phase
incident
when it reaches the same point on the screen:
plane wave E2 (y, t) = E0 sin(Êt + „), with „ given by
" Eq. 10.1 and ” = d sin ◊.11 Superposition
means just adding, so the total E ˛ at position
y on the screen will be

E(y, t) = E1 (y, t) + E2 (y, t)

! = E0 [sin(Êt) + sin(Êt + „)] . (10.5)
!
"
This is a job for the Wonder Identity12
# = ! sin " A B A B
–≠— –+—
sin – + sin — = 2 cos sin .
Figure 10.7: Top: the two-slit setup with L ∫ 2 2
d ≥ ⁄. Bottom: finding ”. (10.6)
We still have some trig to do, but we can Applying this with – = Êt and — = Êt + „,
see what is going to happen without it. Con- A B A B
sider first the case ◊ = 0, that is, the two ≠„ 2Êt + „
E(y, t) = 2E0 cos sin
rays point straight ahead. The two paths are 2 2
A B A B
equal in length, so the waves from the two „ „
= 2E0 cos sin Êt + . (10.7)
slits arrive at the screen in phase and con- 2 2
structively interfere: the screen is bright. As
11
we move up the screen, ” grows from zero. We use the same amplitude for the two waves –
the extra distance traveled by the lower wave is tiny.
When ” = d sin ◊ = ⁄/2, we have destruc- 12
It is obscure, but it’s exactly what we need,
tive interference and the screen is dark. The hence: wonderful. BTW: no need to memorize this
path difference varies sinusoidally with ◊, so – we’ll give it to you if needed.

Copyright 2024 Peter D. Meyers 105

Princeton University Physics 104 Spring 2024

The sine term is just the oscillation of the the physics is contained in the question “By
wave. More
1 2 important to us is the rest, A = how many wavelengths do the path lengths
2E0 cos „2 , which is the amplitude of the differ?”
oscillating total electric field at position y on
the screen. To find the intensity, which we
Example
defined in Ch. 8 as average power per area,
we use Eq. 8.34. In fact, we will use it twice, Light of wavelength ⁄ = 500 nm is incident
first for I0 , the intensity of each slit alone, normally on a pair of very narrow slits with
and then for I, the total intensity: separation d = 2500 nm. The resulting inter-
‘0 c 2 ference pattern is projected onto a screen a
I0 = E ,
2 0 distance L = 1 m away. a) At what distance
A A BB2
‘0 c 2 ‘0 c „ y from the center is the third bright fringe
I= A = 2E0 cos (counting the center fringe as 0)? b) What
2 2 2
A B is the intensity I a distance 5 cm from the
‘0 c 2 „
= E0 4 cos2 center, compared to the intensity I0 from one
2 2 slit alone?
A B
„
∆ I = 4I0 cos2 , (10.8) We can answer part a) by starting with
2 the question two paragraphs above. We
where in the third line we recognize the un- need constructive interference to get a bright
derlined fragment as I0 , the intensity from fringe. The path length difference ” is thus
one slit alone, from the first line. an integer multiple of ⁄. For the 3rd fringe,
” = 3⁄. We then have
To get the intensity vs. y we have some
unpacking to do. We can use 3⁄ = ” = d sin ◊,
” sin ◊ = 3⁄/d = 3(500 nm)/(2500 nm) = 0.6,
„ = 2fi , ” = d sin ◊ (10.9)
⁄ ◊ = 0.64 radians = 36.9¶ ,
to rewrite Eq. 10.8 as y = L tan ◊ = 0.75 m. (10.11)
A B
2 fid sin ◊ For part b), we use ◊ = tan≠1 (y/L) and
I = 4I0 cos . (10.10)
⁄ Eq. 10.10. Try it – I got I ¥ 2I0 . Figure 10.8
shows the interference pattern plotted two
We haven’t substituted to put this in terms ways. Plotted vs. sin ◊, the intensity varies
of y – we will usually settle for intensity vs. ◊, as cos2 ; it is more complicated vs. ◊. Note
or vs. sin ◊. We can then use ◊ = tan≠1 (y/L) that, no matter how you plot it, there are
to find intensity at any y. only five interference maxima on either side
Through all of this, you should always re- of the central maximum, and the fifth one is
member that, no matter how it is presented, at ◊ = 90¶ = fi2 radians.

Copyright 2024 Peter D. Meyers 106

Princeton University Physics 104 Spring 2024

10.4 Multiple narrow

slits
With more than two narrow slits, say n of
them, still with separation d, we have the
situation shown in Fig. 10.9. The same rea-
soning that led to Eq. 10.5 gives

E(y, t) = E1 (y, t) + E2 (y, t) + E3 (y, t)

Figure 10.8: Two-slit interference pattern for + · · · + En (y, t)
the example with d = 5⁄. I/I0 vs. sin ◊ (top) = E0 [sin(Êt) + sin(Êt + „) + sin(Êt + 2„)
and vs. ◊/fi (bottom). The plots span the entire
+ · · · + sin(Êt + (n ≠ 1)„)]. (10.12)
range of angles from ◊ = 0 to ±fi/2.
Can you see that at the center of the screen
OK, but what about this: at the center of (that is, ◊ = 0 æ „ = 0), the intensity with
the screen, I(◊ = 0) = 4I0 . How can the en- n slits is n2 I0 ?
ergy/area/second with two slits be four times
that with one slit and still conserve energy?
as t
The full pattern shows that, since the aver-
&! sin(!" + $) &! increases
age of the function cos2 is 1/2, the average !" + $
intensity is 2I0 , or twice the (uniform) inten-
sity of one slit, as it must be.
Figure 10.10: The “phasor” is a graphical repre-
sentation of a phase. Note that the phasor leads
directly to the sine term like those in Eq. 10.12.
"

! # = ! sin " ! 10.4.1 The phasor diagram

Our Wonder Identity is not applicable to
2# more than two slits; we instead resort to a
graphical technique. While this technique
3#
can be used to compute the complete inten-
4# sity profile, we will use it only to extract a
few important features. We start by noting
that Eq. 10.12 is a sum of sines. The argu-
Figure 10.9: The setup for n narrow slits. ments of the sines are not geometrical angles,
they are phases of an oscillation, but we can

Copyright 2024 Peter D. Meyers 107

Princeton University Physics 104 Spring 2024

still view them in polar form. One term from As time t increases, the arrows in
Eq. 10.12 is shown in Fig. 10.10. The direc- Fig. 10.11a) rotate CCW, keeping the angle
tion of the arrow does not represent that of „ between them fixed. Stare at Fig. 10.11b)
the electric field, or a direction in space at all; and convince yourself that, as time goes on,
it is a phase, so we call the arrow a phasor the two phasors and their resultant A rotate
and not a vector.13 as a unit. The vertical component of A is the
magnitude of E ˛ at the screen. The biggest
In Fig. 10.11a), we show only the first two
this component gets is A, the length of A,
terms of Eq. 10.12, both for simplicity and so
hence the amplitude of the oscillating elec-
that we can compare this to the n = 2 case
tric field at the screen is A.
that we have already solved. Graphically,
taking the sine is finding the vertical com-
ponent of each arrow, and we have to add
them up. Here, phasors add the same way
vectors do: the sum of the vertical compo-
nents is the vertical component of the vector 10.4.2 Examples
sum. This is shown in Fig. 10.11b), where
the resultant A is the “phasor sum” of the
two phasors. Two narrow slits

a) b) The (optional) Appendix to this chapter does

as t
the trigonometry necessary to derive Eq. 10.8
&!
increases
)
$
!" + $
for two narrow slits. We can find the max-
!" + $
!" !" ima and minima (that is, bright and dark
fringes) directly from the phasor diagram.
c) "=0 d) ! = 0 & = 2'! We simplify by setting t = 0, as shown in
$ Fig. 10.11c). We recall, via Eq. 10.9, that „
!= & = 2'!
2 is determined by where we are on the screen
&!
&! $ !=$ &=0
(y or ◊). At the center, y = ◊ = „ = 0.
This situation is the top case in Fig. 10.11d),
Figure 10.11: Phasor diagrams for two narrow where the amplitude of E ˛ is 2E0 and thus
slits. a) Individual phasors. b) Finding the am- I = 4I0 as you saw earlier. The first dark
plitude of E.
˛ c) Situation at t = 0. The simpli- fringe occurs at „ = fi; the two phasors sim-
fied approach for finding maxima and minima ply cancel (bottom of Fig. 10.11d)). We infer
in d) is our main take-away from the phasor that bright fringes occur when „ is a multi-
method. ple of 2fi („ = 0, 2fi, 4fi, . . . ). Dark fringes
occur for odd multiples of fi („ = fi, 3fi, . . . ).
13
We have phasors; Star Trek has phasers. This is the same result we got in Sec. 10.3.

Copyright 2024 Peter D. Meyers 108

Princeton University Physics 104 Spring 2024

Four narrow slits

Intensity/I max
We can use the insight phasors bring us to
find the bright and dark fringes in the case of
four narrow slits. The top case in Fig. 10.12 -1 -0.5 0 0.5 1
shows that „ = an integer multiple of 2fi
sin q
still gives the highest possible intensity and
Figure 10.13: Intensity profiles for 2 narrow slits
thus the brightest fringes. The bottom case,
(orange) and 4 narrow slits (blue). Note that the
„ = fi, still gives a dark fringe, as shown. intensity maxima are set to 1 to compare the
However, it isn’t the only way to get zero in- widths of the peaks. The 4-slit case is actually
tensity. As the middle case shows, „ = fi/2 much brighter.
gives a closed figure and thus a zero resul-
tant A. This is closer to the center bright
fringe than the „ = fi dark fringe is: the cen- of fi do not give dark fringes, just dimmer
tral bright fringe, and all the bright fringes, bright ones. The closed figure in the middle
are narrower than in the two-slit case. Fig- case locates the first dark fringe, and thus
ure 10.13 compares 2- and 4-slit cases with determines how wide the bright fringes are.
the same slit separation d = 6⁄. As found in
our phasor exercise, the 2- and 4-slit bright ! = 0, 2&, 4&, ⋯ * = 3+!
fringes are in the same place on the screen,
but the 4-slit fringes are narrower. 2&
!= *=0
! = 0, 2&, 4&, ⋯ ) = 4*!
3
!=& * = +!
&
!= )=0 Figure 10.14: Phasor diagrams for 3 narrow
2 slits.
!=& )=0
Figure 10.12: Phasor diagrams for 4 narrow
slits. Gratings

The fact that the location of a bright fringe

Three narrow slits depends on ⁄ (it is hidden in „) means that
fringes (other than the central one) of dif-
For an odd number of slits, the bright fringes ferent colors are in different places on the
are still at „ = multiples of 2fi, as shown for 3 screen. We can thus use an array of slits to
slits in the top case in Fig. 10.14. However, observe the spectrum of light from a source,
the bottom case shows that odd multiples that is, its intensity vs. wavelength. As we

Copyright 2024 Peter D. Meyers 109

Princeton University Physics 104 Spring 2024

have seen, the width of a bright fringe de- a sin ◊ = ⁄, then a2 sin ◊ = ⁄2 , and for every
creases as we add more slits of a given spac- slit in the top half of the array, there is a slit
ing. The ultimate version of this is an array in the bottom half that is ⁄2 out of phase and
of thousands of slits called a grating.14 cancels it: the screen is dark at this ◊.
Phasors give the same answer as always
a) b)
for the biggest maxima, that is, the brightest
%
fringes. They occur when all the phasors line
!
sin % =
* up, which happens when „ = m ◊ 2fi with
! = #$ 2 2
m an integer. When talking about gratings,
m is referred to as the “order,” so the first
each !"#$%&'#(
!sin % = * maximum from the center is the “first-order
maximum.” The shortcut here: d sin ◊max =
from previous

Figure 10.15: a) Phasor diagram for the first m⁄ for constructive interference of adjacent
minimum of a grating. b) Simpler construction slits. With a small-angle approximation,
that gives the same result.
◊max = m⁄/d. (10.14)
Figure 10.15a) shows the phasor diagram
for the first dark fringe of a grating with
n slits, and hence the width of a bright fringe.
Intensity/I max

The closed figure is approximately a circle,

taking n steps of „first min to go around 2fi ra-
dians. So
2fi d sin ◊first min 0 0.02 0.04 0.06
= „first min = 2fi
n ⁄ q (rad)
⁄ ⁄ Figure 10.16: Red (⁄ = 650 nm) and orange
sin ◊first min ¥ ◊first min = = , (10.13)
nd a (585 nm) light barely resolved by Rayleigh’s cri-
where we have used Eq. 10.1 for the phase terion using a 10-slit grating with d = 10000 nm.
The central maximum is on the left, and the
difference at the minimum and, in the second
first-order maximum is on the right.
line, we have used a small-angle approxima-
tion, as we will generally be at small angles.
For gratings, the important feature is the
In the last step, we introduce a = nd, the
resolution: how close can two colors (wave-
full width of the array of slits.
lengths) be and still appear separate on
There is an easy-to-remember shortcut to the screen? A rule of thumb, known as
this result, illustrated in Fig. 10.15b). When the “Rayleigh criterion,”15 is that two wave-
14 15
It is actually almost always referred to as a That’s Lord Rayleigh to you and me: John
“diffraction grating.” William Strutt, 3rd Baron Rayleigh.

Copyright 2024 Peter D. Meyers 110

Princeton University Physics 104 Spring 2024

lengths are resolved when, for a given inter- of the whole array fixed at nd = a, the width
ference order m, the maximum of one sits on of the wide slit. This, and a new view of
the first minimum of the other. Figure 10.16 Fig. 10.15, are all we need to find the fea-
makes this plausible. tures most important to us.
We can find an expression for the reso- We start with Fig. 10.15b), where the ver-
lution ⁄ of a grating by taking the angle tical row of dots now is just the dividing
from a maximum to the nearest minimum, points between adjacent bits of the wide slit
◊first min = ⁄/(nd) in Eq. 10.13 and setting of width a. The reasoning used for the grat-
it equal to the angle between the maxima ing still works: when ◊ is such that the path
of two wavelengths differing by ⁄ using length to the screen for the center of the wide
Eq. 10.14, slit is ⁄/2 more than that for the top edge of
the slit, there is destructive interference be-
⁄ m ⁄ ⁄
= ∆ ⁄= . (10.15) tween any point in the top half of the slit and
nd d mn the corresponding point in the bottom half.
We thus have found our first minimum – it
As an example, verify this result for the
is still given by Eq. 10.13. For a grating, the
case in Fig. 10.16.16
next full maximum came when „ = 2fi, but,
for the wide slit we have let n æ Œ, so „,
originally the phase between adjacent pha-
10.5 Diffraction from a sors, never gets that big. Finding the full
wide slit algebraic form of the intensity profile of a
wide slit is beyond the scope of our course;
Up to now, we have been dealing only with for reference, Fig. 10.17 shows an example
narrow slits, that is, slits on the order of a with a = 5⁄. Note that only a narrow re-
wavelength of the incident light or smaller. gion around the center of the screen is illu-
Such a slit gives an outgoing wave that is minated and that the secondary maxima are
approximately uniform in intensity for all ◊. very small. Our interest is primarily in the
As in Fig. 10.4, if the slit is wider, the il- half-width of the central peak, as defined by
luminated part of the screen gets narrower. the location of the first minimum, Eq. 10.13.
For the illuminated part of the screen, there
is still interference, now from different places
in the wide slit. We can treat the wide slit as 10.6 Diffraction-limited
the limit of many narrow slits, letting n æ Œ
as their separation d æ 0, keeping the width resolution
16
That is, show that, for a 10-slit grating in first
order (m = 1), at ⁄ = 650 nm, ⁄ = 65 nm. This is The spread due to diffraction of light passing
(650 ≠ 585) nm, as shown in the Figure. through an aperture is of critical importance

Copyright 2024 Peter D. Meyers 111

Princeton University Physics 104 Spring 2024

ror.
Intensity/I max

-1 -0.5 0 0.5 1
sin q
Figure 10.17: Intensity pattern for one wide slit
with width a = 5⁄. Note the first minimum at
sin ◊ = 0.2.

to optical instruments. This spread, as illus- Example

trated by the width of the central maximum,
is an irreducible smearing of an image in a
telescope, microscope, or camera and a limit
on its ability to resolve close-together ob- Searching for “exoplanets,” planets orbiting
jects. Our result in Eq. 10.13 is for the one- other stars, is very difficult due to the tiny
dimensional situation of a long slit narrow brightness of the exoplanet compared to its
in width. Images formed by telescopes, etc., star. If you can somehow overcome that, to
are two-dimensional. For a circular aperture, see the exoplanet you must still resolve its
the angular separation of the first diffrac- image from that of its star. Let’s determine
tion minimum from the central maximum is how big a telescope you’d need to resolve an
modified by a factor of 1.22,17 and thus the exoplanet (EP) at the same distance from its
“diffraction-limited angular resolution” is, by star as the Earth is from the sun, 1.5◊1011 m,
Rayleigh’s criterion if its star is 20 light years away. (There are
⁄ ⁄ about 100 stars within 20 ly of us.) We will
◊res = (1-dim) æ ◊res = 1.22 (2-dim), use ⁄ = 500 nm, near the middle of the visi-
a a
(10.16) ble spectrum (see Fig. 9.1).
where a is the diameter of the circular aper-
ture and we have used the small-angle ap-
The star-EP angular separation ◊ is the
proximation.18 The “circular aperture” of
star-EP distance divided by its distance from
the telescope is its front lens or main mir-
Earth. We must convert ly to meters. 1 ly
17
Treating 1.22 as ¥ 1 is not a bad approximation. is the distance light travels in one year, and
18
Note that resolution has nothing to do with mag- we know the speed of light, so 20 ly = 20 ◊
nification of the image. More magnification of an un- 3 ◊ 108 m/s ◊ (365 ◊ 24 ◊ 3600) s = 1.9 ◊
resolved image using the same aperture would just
make a bigger image of the overlapping blobs. For 1017 m, and ◊ = 7.9 ◊ 10≠7 radians. Solving
telescopes on the Earth’s surface, fluctuations in the Eq. 10.16 for a, we get 0.8 m: we would need
atmosphere limit resolution for all but the smallest a telescope with a diameter of about a meter.

Copyright 2024 Peter D. Meyers 112

Princeton University Physics 104 Spring 2024

"!
2
" !
"!
2 " ! #!
"
!
#!
Figure 10.18: Phasor construction to find 2-slit
intensity. #!

10.7 Appendix: using

phasors to find 2-slit
intensity vs. ◊
This optional section demonstrates the use
of phasors to find the full intensity pattern
of two narrow slits. Figure 10.18 is a copy of
Fig. 10.11, relabeled to do some trig. The in-
tensity depends on the amplitude of the total
electric field, indicated by A. With the an-
gles defined in Fig. 10.18, we see that A =
2E0 cos –. Starting with the angle labeled „
and going CCW from the dashed line, we see
that „ + 2— = fi. Looking at either of the
right-triangles with green interior angles, we
see that – + — + fi2 = fi, so — = fi2 ≠ –. Sub-
stituting for —, „ = 2– ∆ – = „/2. Thus
A = 2E0 cos „2 , which is the result for the am-
plitude we got using the Wonder Identity in
Eq. 10.6. This approach can be used to find
the full intensity profiles of multiple slits and
wide slits, but we will not push things that
far.

apertures. Telescopes on Earth are made larger and

larger, not to improve resolution, but to collect more
light from distant, dim objects.

Copyright 2024 Peter D. Meyers 113

Appendix A: Core skills
Along with the physical principles and ex- • Understand and be able to explain the
amples that make up each chapter, there are logic of each step in the use of Gauss’s
skills you must acquire to make any of this Law to find E˛ in cases of spherical, cylin-
useful. In this Appendix, we will point out drical, and planar symmetry.
things you should be able to do after reading • Note: When applying Gauss’s Law,
each chapter. you must state (via a sketch) the Gaus-
sian surface you are using and you must
Ch. 1: Electric Charge, Force, and state how you are using the symmetry
Field of E
˛ to do the integrals. Simply writing
equations will not do. (You need not
• Understand what is meant by a vector give a derivation of the E˛ direction from
field like E,
˛ which assigns a vector to
symmetry, but you must indicate the E ˛
every point in space. direction.)
• Be able to visualize a given E, ˛ as in
• Understand the implications of Gauss’s
Figs. 1.4, 1.7, 1.8. Law in the case of conductors in static
• Understand the concept of superposi- electric fields, and be able to use this
tion. to infer how the charge is distributed on
• Understand how we make a sum over the surfaces of the conductors.
charges into an integral over position to
handle continuous charge distributions,
for example Ch. 3: Electric Potential
⁄ ⁄ • Understand the distinction between
ÿ
qi æ dq æ ⁄dx. electrostatic potential and electrostatic
potential energy.
• Perform the change of variables shown • Know and be able to use the potential
in Fig. 1.11. of a point charge.
• Check answers by taking limits of vari- • Understand that there are stwo distinct
ous variables, à la Eqs. 1.14 and 1.25. methods for finding
s
V (˛r) : E˛ · d˛¸
(Eq. 3.4) and k dq/R (Eq. 3.7).
• Be able to get V from E ˛ and E ˛ from V .
Ch. 2: Gauss’s Law and Conductors • Know the meaning of each of the terms
• Picture what happens to make E ˛ = 0 in Q = CV .
inside a conductor when it is placed in • Be able to calculate the capacitance of
an electric field. a pair of conductors.
• Understand the meaning of each term • Have a mental picture of what happens
and symbol in Gauss’s Law, Eq. 2.3. to a dielectric in a static electric field.

114
Princeton University Physics 104 Spring 2024

Ch. 4: Current and DC Circuits the B

˛ direction.)

• Picture current as the net motion of

point charges. Ch. 6: Electromagnetic Induction
• Be able to calculate the resistance of an • Understand what each symbol in Fara-
object given its geometry and resistivity. day’s Law means and be able to identify
• Determine the equivalent resistance of them in various situations.
resistors in series and parallel. • Be able to describe many ways in which
• Apply Kirchhoff’s rules to solve multi- „m can change.
loop DC circuits. • Be able to apply Faraday’s Law to find
• For circuits including capacitors, use induced currents.
Kirchhoff to get a differential equation • Be able to deduce the direction of in-
and then solve it. duced EMF from Lenz’s Law.
• For an RC circuit, find I(0) and I(Œ) • Be able to calculate the inductance of a
without solving the differential equa- device in simple situations.
tion. • Know how to solve circuits containing
• For a given circuit, find the power dissi- inductors.
pated in any component. • For an LR circuit, find I(0) and I(Œ)
without solving the differential equa-
tion.
Ch. 5: Magnetic Forces and Fields • Know how to calculate the energy stored
• Understand the directions involved in in an inductor carrying current.
F
˛ = q ˛v ◊ B.
˛
• Find the magnetic force on moving
charges and currents in wires.
• Use Biot-Savart to calculate the mag-
netic field of a given geometry of cur-
rent.
• Use Ampère’s Law to find B ˛ in cases of
sufficient symmetry.
• Note: When applying Ampère’s Law,
you must state (via a sketch) the
Ampèreian loop you are using and you
must state how you are using the sym-
metry of B ˛ to do the integral. Simply
writing equations will not do. (You need
not give a derivation of the B
˛ direction
from symmetry, but you must indicate

Copyright 2024 Peter D. Meyers 115

Princeton University Physics 104 Spring 2024

Ch. 7: AC Circuits to assign a meaning to it in any given

situation.
• Facility with complex arithmetic.
• Know the impedance of individual com- Needed from (review of) PHY103:
ponents R, L, C. • Understand the structure of the wave
• Be able to find the limiting behavior of equation.
AC circuits at low and high frequencies • Understand the f (x ± vt) structure of
without doing a full calculation. traveling-wave solutions to the wave
• Be able to apply our general procedure equation.
to find the amplitude and phase of I(t) • Know how to parse sinusoidal waves
for combinations of R, L, C using com- given in various forms such as A sin(kz+
plex impedance. Êt) to extract things like amplitude,
• Be able to find voltage across individual wavelength, frequency, and wave speed.
components.
Back to PHY104:
• Be able to calculate the power dissipated
in an AC circuit. • Understand the concept of a plane wave.
• Understand the concept of frequency de- • Understand the properties of EM plane
pendence of amplitudes of current, volt- waves, including the relation among the
age, and power. directions of E,
˛ B,˛ and the direction of
• Specifically, understand the concept of propagation and how E ˛ and B ˛ are re-
“resonance” and be able to find the res- lated.
onant frequency of a circuit that has a • Know how intensity, energy density, and
resonance. power are related and how each is re-
lated to E
˛ and B.˛

Ch. 8: Maxwell’s Equations and EM

Waves
Unlike in the previous chapters, this chap-
ter contains some extensive development
that we do not require you to master. We
want you to follow, appreciate, and, frankly,
enjoy how Maxwell’s Equations combine to
give EM waves. In no way do we ask that
you be able to reproduce the derivation.
• Know how to apply the Maxwell-
Ampère equation.
• Understand what each symbol in each
of Maxwell’s Equations means, and how

Copyright 2024 Peter D. Meyers 116

Princeton University Physics 104 Spring 2024

Ch. 9: Phenomenology of Light as an in Figs. 10.12 and 10.14.

EM Wave • Understand how this results in more
slits giving narrower maxima.
• Be able to compute the effects of an ideal • Know what is meant by “resolution” of
polarizer on unpolarized light and light a grating.
polarized at an angle to the polarizer • Locate the first minimum in intensity of
axis. a single wide slit through the “simpler
• Apply “The angle of reflection equals construction” of Fig. 10.15b).
angle of incidence.” • Find and interpret the angular resolu-
• Use the index of refraction to determine tion of optical instruments with a given
the speed and wavelength of light in a aperture.
transparent medium.
• Apply Snell’s Law to find the direction
of refracted light.
• Understand the circumstance in which
total internal reflection can occur and
calculate the critical angle.

Ch. 10: Interference and Diffraction of

Electromagnetic Waves

• Understand the relation between path-

length difference, phase difference, and
the resulting interference of waves.
• Interpret this for thin films, including
the possible phase change on reflection.
• Be familiar with the notion that, for slits
with width on the order of a wavelength,
a narrower slit illuminates a wider part
of a screen.
• Be able to reproduce the construction of
Fig. 10.7 to relate angle to a point on the
screen to path and phase difference.
• Be comfortable applying the (given)
Wonder Identity to find the full inten-
sity pattern for two narrow slits.
• Be able to use phasors to locate the min-
ima and maxima of multi-slit setups, as

Copyright 2024 Peter D. Meyers 117