Welcome to the archival Web page for U.C. Berkeley's Physics 110A, Section 1, Fall 2001.

Email to:
(Prof.) Mark Strovink, strovink@lbl.gov . I have a research web page, a standardized U.C. Berkeley web
page, and a statement of research interests.

Many of the problem set solutions for this course were handwritten initially by Graduate Student
Instructor Ed Burns. They were refined, typeset, and additional solutions were composed by this course's
GSI Daniel Larson.
Course documents:
Most documents linked here are in PDF format and are intended to be displayed by Adobe Acrobat
[Reader], version 4 or later (Acrobat will do a better job if you uncheck "Use Greek Text Below:" on
File-Preferences-General).
General Information including schedules and rooms.
Course Outline.

Problem Set 1 Solution Set 1

Problem Set 2 Solution Set 2
Problem Set 3 Solution Set 3
Problem Set 4 Solution Set 4
Problem Set 5 Solution Set 5
Problem Set 6 Solution Set 6
Problem Set 7 Solution Set 7
Problem Set 8 Solution Set 8
Problem Set 9 Solution Set 9
Problem Set 10 Solution Set 10
Problem Set 11 Solution Set 11

Midterm 1 Solution to Midterm 1

Midterm 2 Solution to Midterm 2
Final Exam
Mark Strovink
Professor
Particle Experiment

Mark Strovink, Ph.D. 1970 (Princeton). Joined UC Berkeley faculty in 1973 (Professor
since 1980). Elected Fellow of the American Physical Society; served as program advisor
for Fermilab (chair), SLAC (chair), Brookhaven, and the U.S. Department of Energy;
served as D-Zero Physics Coordinator (1997 & 1998).

Research Interests
I am interested in experiments using elementary particles to test discrete
symmetries, absolute predictions and other fundamental tenets of the Standard
Model. Completed examples include early measurement of the parameters
describing charge parity (CP) nonconservation in K meson decay; establishment
of upper limits on the quark charge radius and early observation of the effects of
gluon radiation in deep inelastic muon scattering; and establishment of stringent
limits on right-handed charged currents both in muon decay and in proton-
antiproton collisions, the latter via the search for production of right-handed W
bosons in the D-Zero experiment at Fermilab.
After the discovery in 1995 by CDF and D-Zero of the top quark, we measured
its mass with a combined 3% error, yielding (with other inputs) a stringent test of
loop corrections to the Standard Model and an early hint that the Higgs boson is
light. If a Higgs-like signal is seen, we will need to measure the top quark mass
more than an order of magnitude better in order to determine whether that
signal arises from the SM Higgs.

Current Projects
A continuing objective is to understand better how to measure the top quark
mass. Top quarks are produced mostly in pairs; each decays primarily to b + W.
The b’s appear as jets of hadrons. Each W decays to a pair of jets or to a lepton
and neutrino. For top mass measurement the most important channels are those
in which either one or both of the W’s decay into an electron or muon. For the
single-lepton final states, we developed in 1994-96 and applied in 1997 a new
technique that suppresses backgrounds (mostly from single W production)
without biasing the apparent top mass spectra. For the dilepton final states,
where backgrounds and systematic errors are lower but two final-state neutrinos
are undetected rather than one, a likelihood vs. top mass must be calculated for
each event. During 1993-96 we developed a new prescription for this calculation
that averages over the (unmeasured) neutrino rapidities, and we used it in 1997
to measure the top mass to ~7% accuracy in this more sparsely populated
channel. In both channels, further improvements to measurement technique as
well as accumulation of larger samples will be necessary.
 While studying data from the 1992-1996 CDF and D-Zero samples that contain
both an electron and a muon, we became aware of three events that cannot easily
be attributed either to top quark decay or to backgrounds. Generally this is
because the transverse momenta of the leptons (electrons, muons, and neutrinos
as inferred from transverse momentum imbalance) are unexpectedly large. We
anticipate confirming data e.g. from the D-Zero run that began in 2001.
Transverse momentum imbalance is a broad signature for new physics. For
example, in many supersymmetric models, R-parity conservation requires every
superparticle to decay eventually to a lightest superparticle that, like the
neutrino, can be observed only by measuring a transverse momentum imbalance.
Reliable detection of this signature is one of the severest challenges for collider
detectors. D-Zero’s uniform and highly segmented uranium/liquid argon
calorimeter yields the best performance achieved so far. Building on that, we
have developed a new approach to analysis of transverse momentum imbalance
that, for a given efficiency, yields up to five times fewer false positives.
Recently we have grappled with the long-standing problem of searching with
statistical rigor for new physics in samples that should be describable by
Standard Model processes – when the signatures for new physics are not strictly
predefined. We have identified plausible methods for performing this type of
analysis, and have exercised them on D-Zero data, but the methods involve
sacrifices in sensitivity that we are still working to mitigate.

Selected Publications
S. Abachi et al. (D-Zero Collaboration), “Search for right-handed W bosons and
heavy W′ in proton-antiproton collisions at √s = 1.8 TeV,” Phys. Rev. Lett. 76,
3271 (1996).
S. Abachi et al. (D-Zero Collaboration), “Observation of the top quark,” Phys. Rev.
Lett. 74, 2422 (1995).
B. Abbott et al. (D-Zero Collaboration), “Direct measurement of the top quark
mass,” Phys. Rev. Lett. 79, 1197 (1997); Phys. Rev. D 58, 052001 (1998).
B. Abbott et al. (D-Zero Collaboration), “Measurement of the top quark mass
using dilepton events,” Phys. Rev. Lett. 80, 2063 (1998); Phys. Rev. D 60, 052001
(1999).
V.M. Abazov et al. (D-Zero Collaboration), “A quasi-model-independent search
for new high pT physics at D-Zero,” Phys. Rev. Lett. 86, 3712 (2001); Phys. Rev.
D 62, 092004 (2000); Phys. Rev. D 64, 012004 (2001).

(510) 486-7087 strovink@lbl.gov

 University of California, Berkeley
Physics 110A, Section 1, Fall 2001 (Strovink)

GENERAL INFORMATION (25 Sep 01)

Web site for this course: http://d0lbln.lbl.gov/110af01-web.htm .

Instructors: Prof. Mark Strovink, 437 LeConte; (LBL) 486-7087; (home, before 10) 486-8079; (UC) 642-
9685. Email: strovink@lbl.gov . Web: http://d0lbln.lbl.gov/ . Office hours: M 3:15-4:15,
5:30-6:30.
Mr. Daniel Larson, 281 LeConte, (UC) 642-5647. Email: dtlarson@socrates.berkeley.edu . Office
hours (in 281 LeConte): Th 11:30-12:30, F 2-3.

Lectures: MWF 10:10-11:00 in 329 LeConte, and Tu 5:10-6:30 in 329 LeConte. The Tu 5:10-6:30 slot will
be used occasionally during the semester for the midterm exams; for reviews and special lectures; and for
lectures that substitute for those which would normally be delivered later in the week. Lecture attendance is
strongly encouraged, since the course content is not exactly the same as that of the text.

Discussion Sections: W 4:10-5 in 409 Davis, and Th 4:10-5 in 385 LeConte. Begin in second week. Taught
by Mr. Larson. You are especially encouraged to attend discussion section regularly. There you will learn
techniques of problem solving, with particular application to the assigned exercises.

Texts:
rd
• Griffiths, Introduction to Electrodyamics (3 ed., Prentice-Hall, 1999, required). Get the fourth (or later)
printing, which has fewer typos. I feel that this text is well written and pedagogically effective, though its
scope is modest and its problems are sometimes not very physical.
• If you are planning to attend physics graduate school, it would be smart now to purchase Jackson, Classical
Electrodynamics (3rd ed., Wiley). Optionally, it can be useful in this course.

Problem Sets: A required and most important part of the course. Eleven problem sets are assigned and graded.
Problem sets are due on Fridays at 5 PM, beginning in week 2. Exceptions: no problem set is due in the week
preceding each midterm exam; the problem set that normally would be due on Friday of the week of the second
midterm exam instead is due four days later, on Tuesday of Thanksgiving week (no other problem set is due on
Thanksgiving week). Deposit problem sets in the box labeled “110A Section 1 (Strovink)” in the second floor
breezeway between LeConte and Birge Halls. You are encouraged to attempt all of the problems. Students
who do not do so find it almost impossible to learn the material and to succeed on the examinations. Late
papers will not be graded. Your lowest problem set score will be dropped, in lieu of due date extensions for
any reason. You are encouraged to discuss problems with others in the course, but you must write up your
homework by yourself. (It is straightforward to identify solutions that are written collectively; our policy is to
divide the score among the collectivists.)

Exams: There will be two 80-minute midterm examinations and one 3-hour final examination. Before
confirming your enrollment in this class, please check that its final Exam Group 1 does not conflict with the
Exam Group for any other class in which you are enrolled. Please verify now that you will be available for the
midterm examinations on Tu 16 Oct (in 4 LeConte) and Tu 13 Nov (in 50 Birge), both at 5:10-6:30 PM; and for
the final examination on W 12 Dec, 8-11 AM. Except for unforeseeable emergencies, it will not be possible for
the midterm or final exams to be rescheduled. Passing 110A requires passing the final exam.

Grading: 25% problem sets, 35% midterms, 40% final exam. Departmental regulations call for an A:B:C
distribution in the ratio 2:3:2, with approximately 10-15% D’s or F’s. However, the fraction of D’s or F’s
depends on you; no minimum number need be given.
 Physics 110A, Sec. 1 COURSE OUTLINE Fall 2001 (Strovink)

Week Week Lecture Topic Problem Due 5 PM

No. of... reference Set No. on...
(Griffiths)

1 27-Aug 1.1.5, 1.3.2-1.3.6 Vector and tensor transformations, fundamental theorems

1.4-1.6 Curvilinear coordinates, Dirac delta function, theory of vector fields
2.1-2.2.3 Electrostatic fields, Gauss's law
2 3-Sep LABOR DAY
2.2.4-2.3 Electrostatic potential and boundary conditions
2.4, 2.5.1 Electrostatic work and energy, conductors 1 7-Sep
3 10-Sep 3.1.1-3.1.4 Laplace's and Poisson's equation, simple and relaxation solutions
3.1.5, 3.2.1-3.2.2 Uniqueness of solution, method of images
3.3.1 Separation of variables in Cartesian coordinates 2 14-Sep
4 17-Sep 3.4.2, 3.4.4 Ideal electric dipole and its field
4.1-4.2.1 Forces and torques on electric dipoles; polarization
4.3-4.4.1 Gauss's law in dielectrics, D, linear dielectrics 3 21-Sep
5 24-Sep 4.4.3-4.4.4 Energy in dielectrics, forces on dielectrics
5.1.1-5.1.2 Lorentz force law, particle trajectories in static fields
5.1.3 Current, forces on wires, current densities; charge conservation 4 28-Sep
6 1-Oct 5.2, 5.3.1-5.3.2 Biot-Savart law, divergence of B
5.3.2-5.3.4 Ampere's law and applications, static Maxwell equations
5.4.1-5.4.2 Vector potential, magnetostatic boundary conditions 5 5-Oct
7 8-Oct 5.4.3 Ideal magnetic dipole and its field
6.1.1-6.1.2, 6.1.4 Forces and torques on magnetic dipoles; magnetization
6.3, 6.4.1 Ampere's law in magnetic materials, H, linear magnetic media
8 15-Oct --- TBA
(16-Oct) --- MIDTERM 1 (covers PS 1-5), in 4 LeConte
6.4.2 Ferromagnetism 6 19-Oct
9 22-Oct 7.1 Ohm's law, EMF
7.2.1-7.2.2 Faraday's law
7.2.3-7.2.4 Energy in magnetic fields, inductance 7 26-Oct
10 29-Oct 7.3.1-7.3.3 Maxwell's equations in free space
7.3.5-7.3.6 Maxwell's equations in matter, boundary conditions
10.1 Maxwell's equations for potentials; gauge transformations 8 2-Nov
11 5-Nov 8.1.1-8.1.2 Continuity equation, Poynting's theorem
9.1.1-9.1.2 Wave equation in one dimension, general solution, sinusoidal waves
9.2 EM waves in vacuum, energy and momentum
12 12-Nov --- VETERANS DAY
(13-Nov) --- MIDTERM 2 (covers PS 1-8), in 50 Birge
9.3.1-9.3.2 EM waves in a linear insulator, reflection at normal incidence
14 19-Nov 11.1.1-11.1.2 EM fields of an oscillating electric dipole
11.1.1-11.1.2 Electric dipole radiation and power 9 20-Nov
(22-Nov) --- THANKSGIVING
13 26-Nov 9.1.4 Polarization and angular momentum of EM waves; how to control
9.4.1-9.4.2 EM waves in a conductor, reflection at normal incidence
9.5.1, 9.5.3 EM waves in a coaxial cable 10 30-Nov
15 3-Dec --- Interference and coherence of >1 dipole radiator
--- Radiation pattern from >1 dipole and connection to diffraction (Babinet)
--- TBA 11 7-Dec
16 10-Dec ---
(12-Dec) --- Final exams begin
(12-Dec) 8-11 AM 110A FINAL EXAM (Group 1) (covers PS 1-12)

9/25/2001
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink)
Problem Set 1

1. Griffiths 1.14

2. Griffiths 1.16

3. Griffiths 1.21

4. Griffiths 1.33

5. Griffiths 1.38

6. Griffiths 1.46

7. Griffiths 2.6

8. Griffiths 2.16
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink )

Solution Set 1

1. Griffiths 1.14 Under a rotation, the coordinates y and z transform into ȳ = y cos φ + z sin φ and z̄ =
−y sin φ + z cos φ, so we can invert these equations to find y = ȳ cos φ − z̄ sin φ and z = ȳ sin φ + z̄ cos φ.
Using the chain rule:
∂f ∂f ∂y ∂f ∂z ∂f ∂f
= + = cos φ + sin φ
∂ ȳ ∂y ∂ ȳ ∂z ∂ ȳ ∂y ∂z
∂f ∂f ∂y ∂f ∂z ∂f ∂f
= + = (− sin φ) + cos φ.
∂ z̄ ∂y ∂ z̄ ∂z ∂ z̄ ∂y ∂z
Thus

 
(∇f )y cos φ sin φ (∇f )y
= ⇒ (∇f )i = Rij (∇f )j .
(∇f )z − sin φ cos φ (∇f )z
So ∇f transforms like a vector.

2. Griffiths 1.16 The sketch of this vector field appears in Figure 1.44 of the text. (In fact, a whole discussion
appears in Section 1.5.1.) We want to calculate the divergence of the vector field v = r̂/r2 = (x, y, z)/r3 .
If you calculate the divergence in cartesian coordinates, the formula for the divergence is simple, ∇ · v =
(∂vx /∂x) + (∂vy /∂y) + (∂vz /∂z), however, the derivatives can get a little tiresome. Instead, I’ll calculate the
divergence in spherical coordinates. A general vector field in spherical coordinates looks like v = vr r̂+vθ θ̂+vφ φ̂
and the formula for the divergence is a little complicated (see the inside front cover of Griffiths):

1 ∂ 2 1 ∂ 1 ∂vφ
∇· v = 2
(r vr ) + (sin θvθ ) +
r ∂r r sin θ ∂θ r sin θ ∂φ

However, in our case, vθ = vφ − 0, so the calculation isn’t so bad. In fact, since vr = 1/r2 , ∂(r2 vr )/∂r =
∂(1)/∂r = 0. Thus ∇· v = 0! This is surprising, because the vector field certainly looks like it is diverging
away from the origin. The explanation is that the divergence is zero everywhere except at the origin; at the
origin the above calculation fails because the vector field is undefined there.

3. Griffiths 1.21
∂ ∂ ∂
(a) (A · ∇)B = (Ax + Ay + Az )B
∂x ∂y ∂z
∂Bx ∂Bx ∂Bx ∂By ∂By ∂By ∂Bz ∂Bz ∂Bz
= (Ax + Ay + Az )x̂ + (Ax + Ay + Az )ŷ + (Ax + Ay + Az )ẑ.
∂x ∂y ∂z ∂x ∂y ∂z ∂x ∂y ∂z
This can be expressed more succintly using index notation and the summation convention: [(A · ∇)B]i =
Aj ∂j Bi .
(b) As in an earlier problem, the computation is a little easier in spherical coordinates; however, this time I’ll
use cartesian coordinates for variety. For the x component
 
1 ∂ ∂ ∂ x
[(r̂ · ∇)r̂]x =  x +y +z 
2
x +y +z2 2 ∂x ∂y ∂z x + y2 + z2
2
        
1 1 1 1 1 1 1 1
= x +x − 2x + yx − 3 2y + zx − 3 2z
r r 2 r3 2r 2r
1
x x  2  1 x x
= − 3 x + y2 + z2 = − = 0.
r r r r r r
The other two components are the same, just swapping x for y or z. Thus all three components vanish,
so [(r̂ · ∇)r̂] = 0.

1
 (c) Now take va = x2 x̂ + 3xz 2 ŷ − 2xz ẑ and vb = xy x̂ + 2yz ŷ + 3zx ẑ.
 
2 ∂ 2 ∂ ∂
(va · ∇)vb = x + 3xz − 2xz (xy x̂ + 2yz ŷ + 3xz ẑ
∂x ∂y ∂z
= x2 (y x̂ + 0 ŷ + 3z ẑ) + 3xz 2 (x x̂ + 2z ŷ + 0 ẑ) − 2xz(0 x̂ + 2y ŷ + 3x ẑ)
= x2 (y + 3z 2 ) x̂ + 2xz(3z 2 − 2y) ŷ − 3x2 z ẑ

4. Griffiths 1.33. We want to test Stokes’ Theorem for the function v = (xy) x̂ + (2yz) ŷ + (3zx) ẑ over the
triangular region shown in the figure. First let’s calculate the integral of the curl over the triangle’s area.
∇× v = x̂(0 − 2y) + ŷ(0 − 3z) + ẑ(0 − x) = −2y x̂ − 3z ŷ − x ẑ. Since the path around the outside is going
counterclockwise, the convention is that the area element da = dy dz x̂. Thuse (∇× v) · da = −2y dy dz.
     2
2 2−z 2
1  8
(∇× v) · da = −2y dy dz = −(2 − z)2 dz = (2 − z)3  = − .
0 0 0 3 0 3

Now we calculate the path integral around the boundary of the triangle. v · dl = (xy) dx + (2yz) dy + (3zx) dz.
There are three segments. Take the segment along the y-axis first. Here y goes from 0 to 2, but dx and dz
are zero, and x = y = 0. Thus v · dl = 0 so its integral is zero also. Let the second segment be the slanted
side. Again, x = dx = 0, while z = 2 − y and dz = −dy. In traveling up and left y goes from 2 to 0. So
 0 2 2
v · dl = 2 2yz dy = − 0 2y(2 − y) dy = −(2y 2 − 23 y 3 )0 = − 83 . And the final segment is coming down the
z-axis, where dx = dy = 0 and x = y = 0 so v · dl = 0 and thus there is no contribution to the integral. So the

contributions from all three segments give v · dl = − 83 . Thus in this case we have demonstrated that Stokes’
 
theorem holds, namely S (∇× v) · da = P v · dl.

5. Griffiths 1.38 The divergence theorem tells us that the integral of the divergence over the volume equals the
integral of the vector field over the surface. So we need to compute two integrals in each case.
2 1 ∂ 4
 
(a) ∇·
 (r r̂) =
 2 ∂r (r ) = 4r, where I used equation (1.71). Now, (∇· v1 ) dτ = (4r)r2 dr sin θ dθ dφ =
r
 
dΩ = r4 0 (4π) = 4πR4 . On the other hand, on the surface of the sphere, r = R, and so
R R
4r3 dr
 0  2 
v1 · da = (R r̂) · (R2 sin θ dθ, dφ r̂) = R4 dΩ = 4πR4 . The two integrals agree.

(b) ∇· (1/r2 ) r̂ = r12 ∂r
∂
(r2 r12 ) = 0, so (∇· v2 ) dτ = 0. On the other hand, on the surface of a sphere radius R,
  1 
v2 · da = ( R2 r̂) · (R2 dΩ r̂) = dΩ = 4π. The two integrals don’t agree! The reason is that ∇· v2 = 0

except at the origin, where it becomes infinite. Thus our calculation of (∇· v2 ) dτ is incorrect. We’ll
learn how to fix this using the Dirac delta-function. The correct answer is 4π, whic is what we got the
using the surface integral because that method avoids the problem at the origin.

6. Griffiths 1.46

(a) Since the charge is only at the specific point r we will need to use a delta-function. ρ(r) = qδ 3 (r − r ).
 
The volume integral is then ρ(r) dτ = q δ 3 (r − r ) dτ = q as it should be.
(b) The electric dipole is just two different point charges at different places. ρ(r) = qδ 3 (r − a) − qδ 3 (r).
(c) There is no charge anywhere except where r = R, so ρ(r) = Aδ(r − R). We need to integrate over all
 
space to find A. Q = ρ dτ = Aδ(r − R)4πr2 dr = A4πR2 . So A = 4πR Q
2 . Thus ρ(r) = 4πR2 δ(r − R).
Q

7. Griffiths 2.6 The first thing to notice is that any horizontal components of the electric field will cancel because
of the symmetry of the disk. So the resulting electric field will be in the ẑ direction. Then we just need to
add up the contributions of the z-components due to every point on the disk. For a generic point on the disk
√
located at a distance r from the center, the distance to the point P is r2 + z 2 . Thus the z-component of the
1 dq
E field at P due to that point is 4π 2
0 r +z
2 cos θ where θ is the angle between the line connecting P with the

center of the disk and the line connecting P with the generic element of charge on the disk (see Fig. 1). So we

2
 P

θ
z

Figure 1: Problem 7.

see cos θ = √r2z+z2 . To get the total E-field at P we need to add up these contributions for every little element
of charge on the disk. The amount of charge in a small area element is dq = σrdrdφ. Thus
 R  2π 
1 zσrdr dφ 2πzσ R rdr
E = 2 2 3/2
ẑ = ẑ
0 0 4π0 (r + z ) 4π0 0 (r + z 2 )3/2
2
 
2πzσ  R zσ 1 1
= −(r2 + z 2 )−1/2 ẑ = −√ ẑ.
4π0 0 20 z R + z2
2

Finally, it is always good to check the limiting cases of your results. For R → ∞ the second term vanishes and
we’re left with σ/20 which is the electic field due to an infinite sheet of charge. For z  R at first glance the
two terms in brackets cancel, so we’d get zero, which is true, but not so informative. To get a more precise
estimate of how the field falls off for large z we need to expand the square root.
  −1/2   4 
1 R2 1 R2 R
√ = z2 1 + 2 = 1− 2 +O
2
R +z 2 z z 2z z4

Since z  R, we can ignore the R4 /z 4 piece. Plugging the expansion back into the formula for the electric
σπR2 2
field, we find E → 4π 0z
2 which is the field a distance z away from a point charge with Q = σπR .

8. Griffiths 2.16 We can use Gauss’s law with cylindrical surfaces to determine the electric field in each region.
Following the notation in Griffiths, ŝ is the radial unit vector in cylindrical coordinates.

(i) For s < a, we imagine a small cylindrical surface of radius s and length  inside the inner cylinder. We
know by symmetry that the resulting electric field must point radially outward and is thus perpendicular

to the curved sides of our Gaussian cylinder. Thus E·da = 2πsE(s). Gauss’s law tells us this is equal to
1 1 2
0 Qenc = 0 πs ρ. Equating these two expressions and solving for the electric field we find E(s) = 20 ŝ.
sρ

(ii) Now we imagine our gaussian surface to be between the cylinders. The flux of electric field leaving the

cylinder is the same as above, namely E · da = 2πsE(s), while the other half of Gauss’s law gives
1 1 2 a2 ρ
0 Qenc = 0 πa ρ. Thus for a < s < b, E(s) = 20 s ŝ.

(iii) E · da = 2πsE(s) = 10 Qenc = 0 because we are told that the whole wire is neutral, so the enclosed
charge on the central cylinder in cancelled by the enclosed charge on the outer cylindrical shell. Thus for
s > a, E = 0.
The plot of the electric field as a function of s starts at zero, increases linearly until s = a, then it decreases
along a segment of a hyperbola until s = b, and which point it abruptly drops to zero for all s > a.

3
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink)
Problem Set 2

1. Griffiths 2.18 everywhere within the outer sphere. Construct

a Gaussian surface consisting of a third sphere
2. Griffiths 2.20 at radius r, where R1 < r < R2 . Consider the
volume integral of
3. Griffiths 2.25 (c) only
ρ φ
∇·E− + 2
4. Griffiths 2.32 (a) and (b) only 0 λ̄

5. Griffiths 2.36 (a), (b), and (c) only within that surface. Using the divergence the-
orem, convert it to a surface integral over the
6. Griffiths 2.39 Gaussian surface. Evaluate the integral to ob-
tain the (radial) electric field at r. Your result
7. Griffiths 2.50 should contain a term proportional to the charge
q on the inner sphere, and another small term
8. According to the Proca equations (a rel- proportional to m20 . Integrate this electric field
ativistically invariant linear generalization of from R1 to R2 to solve for the potential dif-
Maxwell’s equations accommodating the pos- ference v that would be measured between the
sibility of a finite rest mass m0 for the photon), inner and outer spheres.
Gauss’s law is modified to become Assuming that q = 0, R1 = 0.5 m, and R2
= 1.5 m, and that V0 is 10 kV, find the volt-
ρ φ
∇·E= − 2 , age v between the inner and outer spheres that
0 λ̄ would be observed if the photon had a rest mass
where φ is the electrostatic potential and m0 = 10−15 eV/c2 .

h̄
λ̄ ≡
m0 c

is the reduced (by 2π) Compton wavelength of

the photon.
Following Williams, Faller, and Hill, Phys. Rev.
Lett. 26, 721 (1971), consider two concentric
spherical perfectly conducting shells of radii R1
and R2 , respectively, with R2 > R1 . Imagine
that the inner sphere is isolated and that the
outer shell is driven by an RF oscillator so that
it has a potential (relative to ∞)

φ2 (t) = V0 cos ωt .

In the modified form of Gauss’s law, make the

following approximation for the value of φ which
appears in the last term; this is a valid ap-
proach because the factor λ̄−2 multiplying it is
very small. The approximation is to set φ = φ2
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink )

Solution Set 2

1. Griffiths 2.18 First we need the electric field inside a uniformly charged sphere. Imagining a spherical Gaussian
surface of radius r inside the charged sphere, symmetry tells us that the electric field must be pointing radially


outward and have the same magnitude over the whole surface. So E(r)· da = 4πr2 E(r) = Qenc /0 = ρ4πr3 /3.
Thus E(r) = ρr/(30 ) r̂ = ρr/(30 ) r where r is the unit vector from the center of the charged sphere to the
point in question.
In this problem we have two charged spheres with the vector d pointing from the center of the positive sphere
to the center of the negative sphere. For a point P in the region of overlap, there will be a contribution to the
E-field from both spheres. If we let r+ be the vector from the center of the positive sphere to P and r− be the
vector from the center of the negative sphere to P , then the total E-field at P is
ρ −ρ ρ
E = E+ + E− = r+ + r− = (r+ − r− ).
30 30 30

But from the figure, we see that r+ − r− = d. So the total electric field in the overlap region is E = ρ/30 d.

−ρ

+ρ d

r−
r+
P

Problem 1

2. Griffiths 2.20 We’ll calculate the curl of E, because a real electrostatic field must have zero curl. For (a) we
have
∇× E = −2kz x̂ − 3kz ŷ − kx ẑ = 0,

so (a) is not a possible electrostatic field. For (b),

∇× E = k(2z − 2z) x̂ + (0 − 0) ŷ + (2y − 2y) ẑ = 0

Thus (b) is a possible electrostatic field. Now we want to compute the potential at some point (x0 , y0 , z0 ),


where the origin is at zero potential, using the relationship that V = − E · d. Let’s choose a simple path
that goes in straight lines from (0, 0, 0) to (x0 , 0, 0) to (x0 , y0 , 0) to (x0 , y0 , z0 ). There are three parts to the
integral, and on each part we have a different E · d = ky 2 dx + k(2xy + z 2 ) dy + 2kyz dz. On the first segment,
y = z = dy = dz = 0 so we get no contribution because E · d = 0. On the second segment, z = dz = dx = 0


y
and x = x0 , so we get the contribution E · d = 2kx0 0 0 y dy = kx0 y02 . On the final segment, dx = dy = 0


z
while x = x0 and y = y0 , so we get E · d = 2ky0 0 0 z dz = ky0 z02 . Now replacing x0 with x and so on,


we find V (x, y, z) = − E · d = −k(xy 2 + yz 2 ). We can check that we did the integrals right by computing
 2 
E = −∇V = k y x̂ + (2xy + z 2 ) ŷ + 2yz ẑ , which gives us back the electric field we started with.

3. Griffiths 2.25 (c) We can use the second equation in (2.30) to calculate the potential due to a disk with
uniform surface charge: 
1 σ(r ) 
V (r) = da .
4π0 r

1
 √
Using cylindrical coordinates with angle φ and radius s, for this case da = σs dsdφ and r = s2 + z 2 .
 
1 2π R
σsds 2πσ  2 R σ  2
V = dφ √ = s + z2 = R + z2 − z
4π0 0 0 s2 + z 2 4π0 0 20
∂V ∂V
Now, V is independent of x and y. Thus ∂x = ∂y = 0, so



∂V σ 1 2z zσ 1 1
E = −∇V = − ẑ = − √ −1 = −√ ẑ.
∂z 20 2 R2 + z 2 20 z R2 + z 2
This is exactly what we got calculating the electric field directly in Problem Set 1.

4. Griffiths 2.32 (a-b) First we need to find the potential and electric field produced by a uniformly charged solid
sphere of radius R and charge q. Outside the sphere, the electric field looks just like that from a point charge,
1 q


E = 4π 0 r
2 r̂. Inside the sphere, we can use Gauss’s Law: E · da = 4πr2 E(r) = Qenc /0 = qr3 /(0 R3 ) ⇒ E =
1 qr
4π0 R3 r̂. To find the potential, we need to do a line integral of the electric field in from infinity. For r > R,
 r
r
1 q  1 q  q 1
V (r) = − dr = = .
∞ 4π0 r2 4π0 r ∞ 4π0 r

For r < R,
 R  r
 
1 q  1 q   q 1 1 r 2 − R2 q 1 r2
V (r) = − dr − r dr = − = 3− 2
∞ 4π0 r2 R 4π0 R3 4π0 R R3 2 4π0 2R R
q
(a) Using equation 2.43, with ρ = inside the sphere and zero outside, we have
(4/3)πR3
   R 
R 2
1 1 3q q r2 2 3q 2 4π 3 r5 1 3q
W = ρV dτ = dΩ 3 − 2 r dr = r − =
2 2 4πR3 0 8π0 R R 2
64π 0 R 4 2
5R 0 4π0 5R

(b) Now using equation 2.45:

 2    R 2 
∞
0 0 2 q 1 2 r 2
W = E dτ = dΩ 4
r dr + 6
r dr
2 2 4π0 R r 0 R

∞
5 R  2
q2 1 r 1 3q
= − + =
8π0 r R 5R6 0 4π0 5R

Fortunately, both the solutions give the same result.

5. Griffiths 2.36 (a-c)

(a) For each of the cavities, we may imagine a Gaussian surface that is completely in the conductor and


surrounds the cavity. Since there is no electric field in the metal of the conductor, E · da = 0. By
Gauss’s Law this means that the charge enclosed must be zero, so the total charge on the inner surface
of the cavity must be exactly opposite of the point charge contained in the cavity. By symmetry, there
is no reason for the surface charge to be anything but uniformly distributed. Thus σa = qa /4πa2 and
σb = qb /4πb2 . Since the conductor is neutral, the charge on the outer surface must be opposite the charge
on the inner surface, and again it will be uniformly distributed, so σR = (qa + qb )/4πR2 .
(b) To find the field outside the conductor, the argument is exactly the same as in Example 2.9 in the text.
The conductor makes the electric field outside look exactly like two point charges qa and qb at the origin.
1 qa + qb
So E = r̂, where r̂ is a unit vector from the center of the conducting sphere.
4π0 r2

2
 (c) The surface charge in cavity a cancels the electric field due to the point charge qa everywhere outside
the cavity. So the only source of electric field in cavity a is the point charge qa and the surface charge.
But using Gauss’s Law and the spherical symmetry, the electric field inside the cavity is just that of the
1 qa
point charge qa , namely Ea = r̂a , where ra is a unit vector from the center of cavity a. The same
4π0 ra2
1 qb
reasoning applies to cavity b, so Eb = r̂b .
4π0 rb2
6. Griffiths 2.39 To find the capacitance between two conductors, we imagine placing a charge +Q on one and
−Q on the other and then calculate the potential difference between them. Then we can find the capacitance
from C = Q/V . (Note that the capacitance should depend only on the physical size of the system and not on
the imaginary charge Q.)
So in this case lets put a charge per unit length +λ on the inner cylinder, and −λ on the outer cylinder.
To calculate the potential difference between the two cylinders, we need to integrate the electric field. Since
the charge is evenly distributed, we can draw a Gaussian cylinder with radius r and length L between the


two conductors. The electric field is pointing radially, and we can find its magnitude: E · da = 2πrLE =
Qenc /0 = λL/0 ⇒ E = λ/2π0 r. The potential difference is then
 b  b 
λ 1 λ b
V (b) − V (a) = − E · d = − dr = − ln .
a 2π0 a r 2π0 a

Since the inner cylinder is at a higher potential (the potential drops in going from a to b), the positive voltage
λ
  2π0 L
between the two conductors is V = V (a) − V (b) = 2π ln ab . Then C = Q/V = λL/V = ln so the
0 ( ab )
2π0
capacitance per unit length is C/L =  b  . Note that this is independent of λ and Q.
ln a

7. Griffiths 2.50 The differential form of Gauss’s Law tells us ∇ · E = ρ/0 . Thus in this case, we find ρ =
0 ∇ · E = 0 a. This is a constant, uniform charge density. So why should the electric field point in the x-
direction and not in the y-direction? In fact, it could, because you find exactly the same charge density for the
fields E = ay ŷ and E = (a/3)r. The point is that the differential equations ∇· E = ρ/0 and ∇× E = 0 are
not sufficient to determine the electric field; boundary conditions are also necessary. It is just like asking for a
function whose derivative is 3. There are many such possibilities: f (x) = 3x, g(x) = 3x + 10, h(y) = 3y + c;
until you know some boundary conditions (such as f (0) = 2), you cannot give a unique answer. Knowing the
field you can determine the charge distribution, but it doesn’t work in reverse: knowing the charge distribution
is not always enough to determine the field.

8. Handout We have two concentric spherical shells, and the outer one is being driven with a potential φ2 (t) =
V0 cos ωt. We make the approximation that the potential between the spheres is φ2 (t) everywhere. This would
be true for the original version of Gauss’s Law (i.e. for a zero mass photon), so call that solution the “original
solution”. The original solution is not an exact solution to the new equations, but since the change in the
equations is very small, the original solution must be very close to the “new solution”. So we will plug in the
“original solution” to the new equation and see how much the “original solution” is modified. The error we
make here is second order in the difference between the “original solution” and the “new solution”, so it can
be ignored for the purposes of determining the sensitivity required to carry out this experiment. We start with
the given differential equation and integrate both sides over a sphere of radius R1 < r < R2 .
 

ρ φ2 ρ φ2
∇· E = − 2 ⇒ ∇· E dτ = − 2 dτ
0 λ̄ 0 λ̄

Now we can compute both sides of this equation separately. By symmetry, we assume that the electric field


is purely radial and has the same magnitude all over our spherical surface. Thus E · dτ = 4πr2 E(r). The


second term on the right hand side is independent of r, and ρ dτ = Qenc . So we find, after substituting in

3
the definition of λ̄,



2 Qenc 4 φ2 q V0 m20 c2
4πr E(r) = − πr3 2 ⇒ E(r) = − r cos ωt r̂.
0 3 λ̄ 4π0 r2 3h̄2
Here Qenc = q, the charge on the inner sphere.
Now, to calculate the potential difference we integrate the electric field.
 R2  R2 
q 1 1 V0 m20 c2 1 
V (R2 ) − V (R1 ) = − E · d = − E(r)dr = − + 2 cos(ωt) R22 − R12
R1 R1 4π0 R2 R1 3h̄ 2

Now lets plug in some numbers. The wording is a little confusing, since V0 is the amplitude of φ2 , it can’t
really be the peak-t0-peak voltage. So I’ll just take V0 = 10 kV and compute the amplitude of the measured
voltage. If you did something slightly different, that fine. The other numbers are q = 0, R1 = 0.5 m, R2 = 1.5
m, and m0 = 10−15 eV/c2 . Be careful about converting all the units. I get ∆V = 8.53 × 10−14 cos ωt volts.
Anything between half or twice this value is acceptable.

4
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink)
Problem Set 3

1. Griffiths 3.1

2. Griffiths 3.4

3. Griffiths 3.6

4. Griffiths 3.9

5. Griffiths 3.10

6. Griffiths 3.14

7. Griffiths 3.15

8. Griffiths 3.43
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink )

Solution Set 3 (compiled by Daniel Larson)

1. Griffiths 3.1 We want to calculate the average potential on the surface of a sphere due to a point charge q
located somewhere within the sphere. Define our coordinates so that the sphere of radius R is centered at the
origin and the point charge lies on the z-axis a distance z from the origin. This calculation is identical to the
one on page 114 of the text, except that z < R, so when it comes time to evaluate the integral we will get a


term (z − R)2 = |z − R| = R − z.
1 q
At any point on the sphere, the potential is V = 4π 0 r
where r2 = R2 + z 2 − 2Rz cos θ (see Figure 3.3 for the
setup, but imagine z < R). We need to calculate
 π
1 q R2 sin θ dθ dφ 1 q
2 2 − 2Rz cos θ 
Vave = √ = z + R
4πR2 4π 0 z 2 + R2 − 2Rz cos θ 4π 0 2Rz 0
1 q 

 1 q 


= z 2 + R2 + 2Rz − z 2 + R2 − 2Rz = (z + R)2 − (z − R)2
4π 0 2Rz 4π 0 2Rz
1 q q
= (z + R − (R − z)) =
4π 0 2Rz 4π 0 R
Note the term R − z as mentioned above, since z < R for a charge inside the sphere. Also notice that the above
result doesn’t depend on the exact location of the point charge. Thus if there were more than one charge, we
Qenc
would find Vave = 4π 0R
. Putting this together with the result in the text for charges outside the sphere, we
have
Qenc
Vave = Vcenter +
4π 0 R
2. Griffiths 3.4 We have a region of space enclosed by one or more boundaries, the charge density ρ is given
inside the region, and either V or ∂V
∂n is specified on each boundary. (The situation is much like Figure 3.6 in
the text, but we’re not assuming any surface is a conductor.) To prove that a solution is unique, we assume
that there are two different solutions and then show that they must be equal. So assume that there are two
different electric fields E1 and E2 in the region that satisfy
ρ ρ
∇· E1 = = −∇2 V1 ∇· E2 = = −∇2 V2
0 0
Now let E3 = E1 − E2 and E3 = −∇V3 = −∇(V1 − V2 ). Subtracting the above equations we find ∇· E3 =
∇· E1 − ∇· E2 = 0. Then

∇· (V3 E3 ) = V3 (∇· E3 ) + E3 · (∇V3 ) = E3 · (−E3 ) = −(E3 )2

Now using the divergence theorem on the above equation for a surface Si that encloses a volume Vi , we have:
 
V3 E3 · da = − (E3 )2 dτ. (1)
Si Vi

Now there are two cases. (I) If the potential is specified on the surface Si , then we must have the two different
potentials agree there, namely V1 (Si ) = V2 (Si ), which means
 
V3 E3 · da = (V1 − V2 )E3 · da = 0.
Si Si

∂V1 ∂V2
(II) If the normal derivative ∂V∂n = ∇V · n̂ is specified on the surface Si , then we must have ∂n (Si ) = ∂n (Si ),
so
     
∂V1 ∂V2
V3 E3 · da = − V3 ∇(V1 − V2 ) · n̂da = − V3 (∇V1 · n̂ − ∇V2 · n̂)da = − V3 − da = 0.
Si Si Si Si ∂n ∂n

1
 But looking back at equation (1) above, we see that both cases imply
 
2
0= (E3 )2 dτ = |E1 − E2 | .
Vi Vi

If we do the integral over all the surfaces in the region, the volume Vi is simply the total volume of the region.
Since the integrand, |E1 − E2 |2 ≥ 0, the only way the above equation can hold is if the integrand is in fact equal
to zero, which means E1 = E2 . Thus the field is uniquely determined if the charge density is given everywhere
and either V or ∂V
∂n is specified on each boundary.

3. Griffiths 3.6 The xy plane is a grounded conductor, so it is at zero potential. We can reproduce this situation
by considering a similar setup without the conductor, but instead with a charge +2q at z = −d and a charge
−q at z = −3d. These image charges make the potential V = 0 anywhere in the xy plane, so it exactly matches
the boundary conditions in the original problem with the conductor. The force on the charge +q is then given
by Coulomb’s Law:
   
1 (q)(−q) (q)(2q) (q)(−2q) 1 29q 2
F= + + ẑ = ẑ.
4π 0 (6d)2 (4d)2 (2d)2 4π 0 72d2

4. Griffiths 3.9 Again, we want to find some image charges that give V = 0 in the xy plane. So we put a uniform
line charge −λ parallel to the x-axis and a distance d directly below it.

(a) The potential due to a single infinite line charge is V (r) = − 4π 2λ
0
ln (r/r0 ) where r is the perpendicular
distance to the line charge and r0 is an arbitrary reference distance. Let’s choose the reference distance to
be d for both the positive and negative line charges; this automatically gives zero potential on the xy-plane.
We want to find the potential at an arbitrary point in the yz plane (the potential must be independent
of x because of translational symmetry in the x-direction). Let s+ and s− be the perpendicular distance
between the point P = (y, z) and the positive and negative line charges. The potential at P is the sum of
the potentials due to each line charge:
 2 
2λ  s− s+  2λ s− λ s2− λ y + (z + d)2
V (y, z) = ln − ln = ln = ln = ln
4π 0 d d 4π 0 s+ 4π 0 s2+ 4π 0 y 2 + (z − d)2

We can check our result by verifying that V (z = 0) = 0 as it must, since the conductor in the xy plane is
grounded.
(b) To find the charge density on the conducting plane of the original problem, we make use of Equation 2.49.
In this case the normal to the xy plane is in the z direction.
   
∂V  ∂V  λ 2(z + d) 2(z − d) 
 λd
σ(y) = − 0  = −  = − − =−
y + (z − d) z=0
0 0 2 2 2 2
∂n z=0 ∂z z=0 4π 0 y + (z + d) π(y 2 + d2 )

5. Griffiths 3.10 We want to find the potential in the first quadrant, so we are only allowed to add image charges
outside this region. We can add an image charge −q at (x, y) = (a, −b) to give zero potential along the x-axis.
To get zero potential along the y-axis we need to add two more image charges to balance the two charges we
have already. They should have opposite charge and be placed as shown in the figure below. Assume all the
charges lie in the xy plane. The potential is the sum of the contributions from the four charges:

1 q q
V (x, y, z) =
+

4π 0 (x − a) + (y − b) + z
2 2 2 (x + a) + (y + b)2 + z 2
2

q q
−
−

(x + a)2 + (y − b)2 + z 2 (x − a)2 + (y + b)2 + z 2

2
 y

−q b +q

x
a

+q −q

Problem 5. Griffiths 3.10

The force on q due to the conducting planes is the same as the force on q due to the image charges, which is
a sum of three contributions. But we need to remember that the force is a vector and keep track of all three
components. First of all, since the charges all lie in the xy plane, there is no z-component: Fz = 0. The other
components follow from Coulomb’s law and breaking the force vectors into components.



1 −q 2 q2 a 1 −q 2 q2 b
Fx = + √ F y = + √
4π 0 4a2 4(a2 + b2 ) a2 + b2 4π 0 4b2 4(a2 + b2 ) a2 + b2

The easiest way to find the work needed to bring the charge q in from infinity into the corner made by the
conducting planes is to compute the total work needed to bring together the collection of image charges and
then divide by 4, because we don’t count the work needed to bring in the image charges, for in the original
problem the only other charges present are those induced in the conductors, but the induced charge comes “for
free” because conductors are equipotential surfaces. Thus the work to bring in the single charge q is (using
equation 2.40 and multiplying by 1/4):
 2   
1 1 q q2 q2 q2 q2 q2 q2 1 1 1
W = − − + √ − − + √ =− + −√
4 4π 0 2a 2b 2 a2 + b2 2a 2b 2 a2 + b2 16π 0 a b a2 + b2
This method would work for any angle which evenly divides 360◦ , namely 360◦ /2n for n = 1, 2, 3, . . ..

6. Griffiths 3.14

(a) In this problem, the pipe is infinite in the z-direction, so there can be no dependence on z because of the
symmetry. Thus we are left with solving Laplace’s equation in two dimensions:

∂2V ∂2V
∇2 V = + =0
∂x2 ∂y 2

Using separation of variables, we assume V (x, y) = X(x)Y (y). Plugging into the above equation, and
letting primes denote derivatives of single variable functions with respect to their argument, we get
1  1
Y (y)X  (x) + X(x)Y  (y) = 0 ⇒ X (x) + Y  (y) = 0
X Y
In order fo this equation to hold for all x and y, we must have both terms equal to constants. Since the
potential must vanish at y = 0 and y = a, it makes sense to use sines and cosines in the y-direction, which
means we want to put a negative constant in the y-equation.
1  1 
Y (y) = −k 2 X (x) = k 2 for k constant
Y X
The solutions for the y equation give Y (y) = A sin ky + B cos ky. For the x equation, we need it to vanish
at x = 0, so lets choose hyperbolic trig functions instead of exponentials: X(x) = C sinh kx + D cosh kx.

3
 Thus V (x, y) = (A sin ky + B cos ky)(C sinh kx + D cosh kx). Now we need to choose the coefficients
A, B, C, D to satisfy the boundary conditions. V (x, y = 0) = 0 means we need B = 0. V (x = 0, y) = 0
means we need D = 0. V (x, y = a) = 0 means we need sin ka = 0 ⇒ ka = nπ for n = 1, 2, 3, . . .. The
most general solution at this stage is a linear combination of solutions for different n, where I’ve combined
the constants A and C into An :
∞  nπx   nπy 
V (x, y, z) = An sinh sin
n=1
a a
∞    
To determine the An we need the last boundary condition, V (x = b, y) = n=1 An sinh nπb sin nπy =
 mπy  a a
V0 (y). Using Fourier’s trick, we multiply both sides by sin a and integrate from 0 to a. This gives
    mπy   ∞  nπy 
nπb 2 a 2
Am sinh = V0 (y) sin dy ⇒ An = V0 (y) sin dy
a a 0 a a sinh(nπb/a) 0 a
Thus the equation for V (x, y, z) together with the formula for An gives a general formula for the potential
within the pipe.
(b) With V0 (y) = V0 = constant, we find
 ∞ 
2  nπy  2V0 0, if n is even,
An = V0 sin dy = 2a
a sinh(nπb/a) 0 a a sinh(nπb/a) nπ , if n is odd.

So we find 
4V0 sinh(nπx/a) sin(nπy/a)
V (x, y) = .
π n=1,3,5,...
n sinh(nπb/a)

7. Griffiths 3.15 Another problem where we need to use separation of variables, but this time with all three
dimensions. Proceeding as before, we assume V (x, y, z) = X(x)Y (y)Z(z) and plug this into Laplace’s equation,
to find
1  1 1
X (x) + Y  (y) + Z  (z) = 0.
X Y Z
Each of these terms must be constant, and the sum of the three constants must be zero. We want to choose the
constants appropriately by looking at the boundary conditions. In the x and y directions there are grounded
plates at 0 and a, which means the solutions will be sines and cosines in those directions, so we choose the
constants for the X and Y terms to be negative. In order to add to zero, the other constant must be positive.
1  1  1 
X (x) = −k 2 Y (y) = −l2 Z (z) = k 2 + l2 for k, l constants
X Y Z
Now we can write down the general solutions to these equations. Since Z must vanish at z = 0, it is easier to
write down the solution in terms of hyperbolic trig functions instead of real exponentials.



X(x) = A sin kx + B cos kx, Y (y) = C sin ly + D cos ly, Z(z) = E sinh(z k 2 + l2 ) + F cosh(z k 2 + l2 )

The boundary conditions tell us V (0, y, z) = V (x, 0, z) = V (x, y, 0), so to make this hold for all values of the
other variables, we must have B = D = F = 0. Then V (a, y, z) = V (x, a, z) = 0 requires k = nπ/a and
l = mπ/a for positive integers n and m. So at this stage, the most general solution is a linear combination of
solutions for all n and m.
 √ 
∞  ∞  nπx   mπy  πz n2 + m2
V (x, y, z) = An,m sin sin sinh
n=1 m=1
a a a

The only thing that remains is to fix the constants An,m by using the last boundary condition: V (x, y, a) = V0 .
Using Fourier’s trick, we set z = a, multiply both sides by a2 sin(n πx/a) a2 sin(m πy/a) and integrate over both
x and y from 0 to a. This will pick out the coefficient An
,m
.
 2  a  a       

2 n πx m πy 0, if n or m is even,
An
,m
sinh(π n2 + m2 ) = V0 sin sin dx dy = 16V 0
a 0 0 a a π 2 n
m
, if both are odd.

4
 The above equation gives us An,m , which we can plug into the double sum above. The final solution is
16V0   1  nπx   mπy  sinh(πz √n2 + m2 /a)
V (x, y, z) = 2 sin sin √ .
π nm a a sinh(π n2 + m2 )
odd n odd m

8. Griffiths 3.43

(a) To use the hint, we need to figure out what it means to integrate by parts in three dimensions. We can
work it out starting from vector identity (5) in the front cover of Griffiths. With a scalar function V and
a vector function E the identity can be written

E · (∇V ) = ∇· (V E) − V (∇· E).

Now we integrate both sides over a volume V with surface S and use the divergence theorem on the first
term on the right hand side. This yields a formula for three-dimensional integration by parts:
  
E · (∇V ) dτ = V E · da − V (∇· E) dτ
V S V

Now assume we have two completely different systems, numbered 1 and 2, each of which has a certain
charge density ρ, potential V and electric field E. Following the hint we will integrate E1 · E2 in two ways.
     
E1 · E2 dτ = − (∇V1 ) · E2 dτ = − V1 E2 · da + V1 (∇· E2 ) dτ = − V1 E2 · da + V1 ρ2 / 0 dτ
V V S V S V

If we assume that the charge distributions are localized (i.e. do not extend to infinity) then we can take
our volume to be all of space, which means that the surface S is at infinity, where the potential V1 falls
off to zero. So the surface integral vanishes, leaving
 
1
E1 · E2 dτ = V1 ρ2 dτ
V 0 V
We can do exactly the same manipulations after replacing E2 with −∇V2 , so we’ll arrive at the same
result with the labels 1 and 2 switched. So we conclude
  
0 E1 · E2 dτ = V1 ρ2 dτ = V2 ρ1 dτ
V V V

(b) Now we want to apply the above result to a specific situation. It will be less confusing if I call the
conductors a and b instead of 1 and 2. In the first system we have two conductors and we put a charge
Q on conductor a, and let Vab be the potential at conductor b. So in this system ρ1 is zero everywhere
except on conductor a, where there is total charge Q distributed in some complicated way. But this means

that V ρ1 dτ = Q. The potential in this system is complicated. The only place we know what it is is on
conductor b, where V1 is a constant, V1 = Vab .
Now consider the second system. It consists of the same two conductors a and b in the same positions,
but this time we put charge Q on conductor b and call the potential at conductor a Vba . Here, ρ2 is zero

everywhere except on conductor b. But we know V ρ2 dτ = Q. The potential is complicated, and all we
know is that on conductor a it is constant and equal to Vba .

Now we apply Green’s reciprocity theorem. When we calculate ρ1 V2 dτ , ρ1 vanishes everywhere except
conductor a, but that is exactly where we know what V2 is; it’s a constant equal to Vba . Thus
 
ρ1 V2 dτ = Vba ρ1 dτ = Vba Q.

But we also have  

ρ2 V1 dτ = Vab ρ2 dτ = Vab Q,

where again we could do the integral because ρ2 is zero everywhere except on conductor b where V1 = Vab .
The Q’s cancel, leaving us with the result Vab = Vba , or in the notation of the problem, V12 = V21 .

5
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink)
Problem Set 4

1. Griffiths 3.33

2. Griffiths 4.4

3. Griffiths 4.6

4. Griffiths 4.10

5. Griffiths 4.13. [Hint: Consider the uniformly

polarized cylinder to be the superposition of two
cylinders that are uniformly charged throughout
their volume, one positive, the other negative,
with a small relative offset. Take the limit as the
offset vanishes while the product of offset and
charge remains finite.]

6. Griffiths 4.15

7. Griffiths 4.16

8. Griffiths 4.18
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink )

Solution Set 4 (compiled by Daniel Larson)

1. Griffiths 3.33 To get a general formula for the electric field from an electric dipole, let’s start with the general
formula for the potential, (3.99) in the text.

 p · r 

1 p · r̂ 1 1 1 1
Edip (r) = −∇Vdip (r) = − ∇ =− ∇ 3 =− ∇(p · r) + (p · r)∇ 3
4π0 r2 4π0 r 4π0 r3 r
To evaluate the first term we use some vector identities.

∇(p · r) = p × (∇× r) + r × (∇× p) + (p · ∇)r + (r · ∇)p = (p · ∇)r,

because ∇× r = 0 and p is a constant vector, so any derivatives of it vanish. To evaluate the one remaining
term we can temporarily choose cartesian coordinates:


∂ ∂ ∂
(p · ∇)r = px + py + pz (x x̂ + y ŷ + z ẑ) = px x̂ + py ŷ + pz ẑ = p.
∂x ∂y ∂z
For the second term we need ∇rn = nrn−1 r̂. Putting the results together,

 
1 1 −3 1 1
Edip (r) = − p + (p · r̂)r r̂ = [3(p · r̂) r̂ − p]
4π0 r3 r4 4π0 r3

2. Griffiths 4.4 The point charge produces an electric field with magnitude E1 = q/4π0 r2 at the location of
the neutral atom. That electric field polarizes the atom, giving it a dipole moment p = αE1 = −αq/4π0 r2 r̂
where r is the vector pointing from the atom towards the point charge. But the polarized atom produces its
own field due to its dipole moment p. At the location of the point charge, the electric field is E2 = 2p/4π0 r3
where I’ve used the result of the previous problem. Finally, the force felt by the point charge is attractive:

2
αq 2 q 1
F = qE2 = − 2 2 5 r̂ = −2α r̂.
8π 0 r 4π0 r5

3. Griffiths 4.6 To determine the effect of the conducting plane, we can use an image dipole situated below the
plane. To figure out how it should be pointing, we can think of the perfect dipole as two charges separated by
a small distance, figure out where the image charges should be, and then let the distance between the charges
in each dipole go to zero. This is shown in the figure.

p θ+ ^r
−
z
p θ ^θ
z
θ θ 2z
+
p − p
i
i

The image dipole, pi , creates an electric field Ei at the position of the real dipole, which causes a torque on
the real dipole, N = p × Ei . If we choose a coordinate system centered on the image dipole with pi pointing
in the z-direction, then the real dipole can be taken to be in the xz-plane with coordinates (r, θ, φ) = (2z, θ, 0).
Using equation (3.103) in the text, the electric field there is E1 = 4π0p(2z)
i
3 (2 cos θ r̂ + sin θ θ̂). Now, in order to

take the cross product, we need to resolve p in the r̂ and θ̂ directions. From the figure we see that p makes
an angle θ with the r̂ direction. Thus p = p cos θ r̂ + p sin θ θ̂.
pi −p2 cos θ sin θ 1 p2 sin(2θ)
N = p × Ei = (p cos θ r̂ + p sin θ θ̂) × 3
(2 cos θ r̂ + sin θ θ̂) = 3
φ̂ = − φ̂
4π0 (2z) 4π0 (2z) 4π0 16z 3

1
 where we have used r̂ × θ̂ = φ̂ and pi = p. Note that −φ̂ is the direction out of the page. The torque vanishes
for θ = 0, π/2, and π. However, since the dipole wants to rotate one way for 0 < θ < π/2 and the other way
for π/2 < θ < π, at π/2 the torque is changing sign and so the dipole is not stable at that angle. The stable
orientations are for θ = 0 or π where the dipole is perpendicular to the conducting plane, pointing either toward
or away from it.

4. Griffiths 4.10
∂
(a) σb = P(R) · n̂ = kR r̂ · r̂ = kR. ρb = −∇· P = − r12 ∂r (r2 kr) = − r12 3kr2 = −3k.
(b) There is no free charge specified, and the sphere is not connected to any wires or batteries, so the
free charge is zero everywhere. Thus the only charges contributing to the electric field are the bound
charges. Because of the symmetry we know that the electric field can only be pointing radially, so
we can use Gauss’s law to find the field. Inside the sphere we make a gaussian sphere of radius r:
E(r)4πr2 = 43 πr3 ρb /0 ⇒ Ein (r) = ρb r/30 r̂ = −k/0 r. Outside the sphere the total volume charge is
−3k 43 πR3 = −4πR3 k while the total surface charge is kR × 4πR2 = 4πR3 k. Thus the net charge inside a
Gaussian surface with radius r > R is zero, which means Eout = 0.

5. Griffiths 4.13 We want to tackle this problem in exactly the same way we did the sphere with uniform
polarization in class, or in example 4.3 in the text. We can think of the uniformly polarized cylinder as two
cylinders with opposite uniform charge density ±ρ separated from each other by a small distance d. Start by
considering a single, uniformly charged cylinder. Using Gauss’s law we can find the electric field both outside
and inside the cylinder. Using Griffiths’s notation with s as the radial coordinate, inside the cylinder we find:
E2πs = 10 ρπs2  ⇒ E = (ρ/20 )s. In the region of overlap between the two cylinders we have contributions to
E from both cylinders which add just like in Problem 2.18 (see solution set 2 for a figure). However, in this case
we’ll define d to be the vector pointing from the center of the negative cylinder to the center of the positive
cylinder. Thus the total electric field in the region of overlap is E = −(ρ/20 )d. We can think of the two
uniformly charged cylinders as being line charges with charge per length ±λ = ±πa2 ρ, which is like a bunch
of dipoles λdd all in a row. Now thinking back to the single, polarized cylinder, the total dipole moment in
a piece of length  is P(πa2 ) = λd = πa2 ρd, so P = ρd. Plugging this into our expression for E we find
Ein = − 210 P.
Now we need the electric field outside the cylinders. This time, for a single uniformly charged cylinder and
s > a Gauss’s law gives: E2πs = 10 πa2 ρ ⇒ E = (ρa2 /20 s) ŝ. At some point outside the cylinders, let s+
and s− be the radial vectors from the centers of the two charged cylinders to the point in question. The total
electric field at that point gets contributions from both cylinders, so



ρa2 ŝ+ ŝ− ρa2 s+ s−
Eout = E+ + E− = − = − 2
20 s+ s− 20 s2+ s−
We want to simplify this expression, using the fact that s+ − s− = −d and d  s+ , s− . Let s be the radial
vector from the midpoint between the two charged cylinders; this is the true center of the uniformly polarized
cylinder. Then s± = s ∓ d2 .


−1

−1
s± d d2 d 1 s·d d2
= s∓ s2 ∓ s · d + = s∓ 1 ∓ +
s2± 2 4 2 s2 s2 4s2




1 d s·d 1 s·d d
≈ s∓ 1± 2 = 2 s±s 2 ∓
s2 2 s s s 2
where we have kept only the terms linear in the small quantity d/s. Using this result in the expression for the
electric field, and the result that P = ρd, we find




ρa2 1 s·d d s·d d ρa2 1 2s(s · d) a2
Eout = s + s − − s − s + = − d = [2(P · ŝ) ŝ − P]
20 s2 s2 2 s2 2 20 s2 s2 20 s2

2
6. Griffiths 4.15
(a)

 
1 ∂ k k P · r̂ = k/b (at r = b),
ρb = −∇· P = − 2 r2 = − 2; σb = P · n̂ =
r ∂r r r P · (− r̂) = −k/a (at r = a).

1 Qenc
The spherical symmetry again tells us that E is radial. Thus E = 4π 0 r
2 r̂ in all three regions. For
a gaussian surface with r < a there is no charge enclosed, so E(r < a) = 0. For a < r < b the charge

r  −k
enclosed is −k 2 
a 4πa + a r
2 4πr dr = −4πka − 4πk(r − a) = −4πkr, so E(a < r < b) = −(k/0 r) r̂.
2

Finally, for r > b the charge inclosed is the same as in the previous calculation (with r = b) plus the
surface charge at r = b. So Qenc = −4πkb + 4πb2 (k/b) = 0, thus E(r > b) = 0.

(b) The spherical symmetry tells us D must be radial, so D · da = 4πr2 D(r) at some radius r. But since
there is no free charge anywhere, we must have bf d = 0 everywhere. Since 0 E = D − P = −P, since
P = 0 both inside and outside the shell, E = 0 both outside and inside the shell. Within the shell,
E = −P/0 = −(k/0 r) r̂. This agrees with part (a) and was far quicker.

7. Griffiths 4.16 We want to find D and E inside the cavity. This is easiest to do by considering the superposition
of the original piece of polarized dielectric without a hole and a piece of dielectric in the shape of the cavity
possessing opposite polarization. The the fields at the center of the cavity will be the sum of the fields due to
the original dielectric (namely E0 and D0 ) with the fields at the center of uniformly polarized objects in the
shape of the cavity, which we can call E and D . It is the latter fields that we must determine.

(a) The fields at the center of a uniformly polarized sphere were found in example 4.3. If P is the polarization
of the original dielectric with a cavity, then −P is the polarization of the cavity-shaped piece we are
superimposing. So E = − 310 (−P). Thus the polarization in the cavity is E = E0 + E = E0 + 310 P.
Also, since the polarization in the cavity is zero, we have D = 0 E = 0 E0 + 13 P = D0 −P+ 13 P = D0 − 23 P.
(b) A long thin needle with polarization −P looks like a bunch of dipoles sitting end to end in a long line,
like Figure 4.11 in the text. Thus the net charge that contributes to the electric field at the center of
the needle are positive and negative charges on the ends of the needle. But for a very long and very thin
needle these will be small charges and far away, so will have negligible contribution. Thus E = 0, so
E = E0 . Again, in the cavity there is no polarization so D = 0 E = 0 E0 = D0 − P.
(c) The thin wafer shape has the field of a parallel plate capacitor with charge σb = P · n̂ = −P on the upper
plate and the opposite charge on the bottom plate. The electric field2 between the plates is then pointing
up, in the same direction as P, and has magnitude P/0 . Thus E = 10 P, so E = E0 + E = E0 + 10 .
Finally, D = 0 E = 0 E0 + P = D0 .

8. Griffiths 4.18 Choose coordinates so that the capacitor is in the xy-plane and ẑ points “up” from the
negatively charged plate towards the positively charged plate. Let’s start this problem by thinking physically
about what will happen. There is free charge placed on the top and bottom plates, which will produce some
electric field pointing down. (We will assume the capacitor is big enough in the xy-directions so that the
electric field will be only in the z-direction.) But that electric field will polarize the two dielectrics, producing
P pointing in the same direction as E, which in turn induces positive bound surface charge on the bottoms
of each dielectric surface and negative bound charge on the top of each dielectric surface. These collections
of bound charge will also produce their own electric field, which we also need to take into account. Now let’s
work through the details.

(a) The D field depends only on the free charge, so with +σ on the top plate and −σ on the bottom plate,
the D field in between the plates will be D = σ(− ẑ), which is the D field between two infinite planes
with free surface charges ±σ. It has the same value in each of the slabs.

3
(b) Since we’re dealing with linear dielectrics, D = 0 r E. However, r is different in the two slabs. Thus in
(1) (2)
slab 1, E1 = D/0 r = −σ/20 ẑ, and in slab 2, E2 = D/0 r = −σ/1.50 ẑ = −2σ/30 ẑ.
(c) In a linear dielectric, P = 0 (r − 1)E, so in slab 1 we have P1 = 0 (2 − 1)E1 = −σ/2 ẑ and in slab 2 we
have P2 = 0 (1.5 − 1)E2 = −σ/3 ẑ.
(d) We find the potential difference by integrating E · d between the two plates. Since E is uniform in each
of the slabs, and points straight down, we get V = E1 a + E2 a = 7aσ/60 .
(e) ρb = −∇ · P = 0 in both slabs. σb = P · n̂, so remembering that n̂ always points out of each slab,
(1) (2)
σb = +σ/2 at the bottom of slab 1 and minus that at the top of slab 1. σb = +σ/3 at the bottom of
slab 2 and minus that at the top. See the figure below.
(f) Using all of the charges, we want to recalculate the electric field in each slab. All of the charges are surface
charges distributed on (approximately) infinite planes of charge, so we use the result that the electric field
due to a single plane of surface charge doesn’t depend on the distance from the plane. Inside slab 1 it is
as if there was a single plane on top with net surface charge equal to the free charge on the top capacitor
plate plus the bound charge on the top of slab 1, namely σ − σ/2 = σ/2; and also a single plane below with
net surface charge dues to the bound charge on the bottom of slab 1, the top of slab 2, and the bottom of
slab 2, and the free charge on the bottom capacitor plate, namely σ/2 − σ/3 + σ/3 − σ = −σ/2. So inside
slab 1 the electric field is the same as between two infinite plates with surface charge ±σ/2, so E1 = σ/20
(pointing down). Similarly, inside slab 2 there is net surface charge σ − σ/2 + σ/2 − σ/3 = 2σ/3 above
and σ/3 − σ = −2σ/3 below. So the electric field in slab 2 is E2 = 2σ/30 , again pointing down.
+σ

−σ/2
Slab 1
+σ/2

−σ/3
Slab 2 +σ/3

−σ
Problem 8. Griffiths 3.18

4
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink)
Problem Set 5

1. Griffiths 5.1

2. Griffiths 5.7

3. Griffiths 5.8

4. Griffiths 5.11

5. Griffiths 5.12

6. Griffiths 5.17

7. Griffiths 5.20

8. Griffiths 5.21
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink )

Solution Set 5 (compiled by Daniel Larson)

1. Griffiths 5.1 Since the field is pointing into the page, a positive charge would feel a force in the direction
v × B, which is up. So the charge is positive. From example 5.1 we know that momentum is p = QBR where
R is the radius of the circle traced out by the charge. Using the pythagorean theorem, and in the figure below,
2
+d2
we find (R − d)2 + a2 = R2 ⇒ R2 − 2Rd + d2 + a2 = R2 ⇒ R = (a2 + d2 )/2d. Thus p = QB a 2d .

R
R

d d
a

Problem 1. Griffiths 5.1

2. Griffiths 5.7 First lets calculate the time derivative of the total dipole moment. Recall the definition of


p = V ρr dτ .
    
dp d ∂ρ
= ρr dτ = r dτ = − (∇· J)r dτ,
dt dt V V ∂t V
where in the last equality we’ve used the continuity equation. Now we need to use the hint, so calculate
∇· (xJ) = x(∇· J) + J · (∇x) = x(∇· J) + Jx because ∇x = x̂. We are assuming that the current is completely
within the volume V, so that means there can be no current leaving through the surface S. But that is true
only if J · da = 0 on S.
     
x(∇· J) dτ + Jx dτ = ∇· (xJ) dτ = xJ · da = 0 ⇒ Jx dτ = − (∇· J)x dτ.
V V V S V V

We can make the same argument with x replaced by y or z. Putting the three results together gives the vector




equation − V (∇· J)r dτ = V J dτ . Combining this with the previous computation, we find dpdt = V J dτ .

3. Griffiths 5.8

(a) We can use the intermediate result from example 5.5, namely equation (5.35). In this case we have
◦
s = R and −θ
√ 1
= θ√2 =
 45 . We also have four such contributions, one from each side of the square. So
2 − 2
√
B = 4 4πR 2 − 2
µ0 I
= µ0 I 2/πR.
(b) Generalizing the previous result, s = R, −θ1 = θ2 = π/n, so B = n 4πR
µ0 I
[sin(π/n) − sin(−π/n)] =
nµ0 I
2πR sin(π/n).
sin(π/n) π/n
(c) Now taking n → ∞, for small x, sin x ≈ x, so for large n we have 1/n ≈ 1/n = π. Thus B = µ0 I/4R,
which is the result in equation (5.38) with z = 0.

4. Griffiths 5.11 We imagine the solenoid to be a series of n circular coils per unit length, each contributing a
a2
field B = µ20 I (a2 +z 2 )3/2 to to point at P , where z is the disance along the solenoid’s axis between P and the

center of the coil (Equation 5.38). To add up the contributions from all the rings, we note that the amount of
current flowing in a section of width dz is nIdz, so we need to integrate over z = a cot θ from one end of the
solenoid to the other.
    
µ0 Ina2 dz µ0 In θ2 a2 −a dθ µ0 nI θ2 µ0 nI
B= 2 2 3/2
= 3 2 3/2 2 = − sin θ dθ = (cos θ2 − cos θ1 )
2(a + z ) 2 θ1 a (1 + cot θ) sin θ 2 θ1 2

1
 For an infinite solenoid, θ2 = 0 and θ1 = π, so cos θ2 − cos θ1 = 1 − (−1) = 2. Hence B = µ0 nI.

5. Griffiths 5.12 Using equation (5.37), the magnetic force of attraction per unit length between two wires
µ0 I1 I2 µ0 λ2 v 2
carrying currents I1 and I2 is fm = 2π d . Since the current in each wire is I = λv, we have fm = 2π d . The
λ2
electric field of one wire at a distance d is E = 2πλ0 d , so the electric repulsion per unit length is fe = λE = 2π 0d
.
2 2
The forces will balance when fm = fe ⇒ µ0 v = 1/0 ⇒ v = 1/0 µ0 . But recall that these fundamental
constants are related to the speed of light: µ0 0 c2 = 1. Thus the forces will balance when v = c = 3 × 108
m/s. Obviously one could never accelerate any physical wires to the speed of light; thus the electric repulsion
always dominates.

6. Griffiths 5.17 Let’s choose coordinates so that the z-axis runs along the axis of the solenoid. We want to
find the magnetic field at any arbitrary point. But we can choose coordinates so that this point is on the
y-axis: r = (0, y, 0). Now we want to look at contributions to the magnetic field from small pieces of current
loops, one above and one below r, as shown in the figure. First consider the contribution from a section on
loop 1 at position (x , y  , z  ), located above r. dl = dx x̂ + dy  ŷ. Also, the vector pointing from (x , y  , z  ) to
r = (0, y, 0) is: r̃ = −x x̂ + (y − y  ) ŷ − z  ẑ. Thus dl × r̃ = (−z  dy  ) x̂ + (z  dx ) ŷ + [(y − y  ) dx + x dy  ] ẑ. So
the contribution to the magnetic field from this piece of loop 1 is:
µ0 I dl × r̃ µ0 I (−z  dy  ) x̂ + (z  dx ) ŷ + [(y − y  ) dx + x dy  ] ẑ
dB1 = =
4π r̃3 4π [(x )2 + (y − y  )2 + (z  )2 ]
3/2

Now we want to consider the contribution from a section of a coil that is at the same position but below r, i.e.
at (x , y  , −z  ). The only difference is that z  changes sign, so the contribution to the magnetic field will be
µ0 I dl × r̃2 µ0 I (z  dy  ) x̂ + (−z  dx ) ŷ + [(y − y  ) dx + x dy  ] ẑ
dB2 = 3 = 3/2
4π r̃2 4π [(x )2 + (y − y  )2 + (−z  )2 ]
When these two contributions are added, the x̂ and ŷ components exactly cancel, leaving only a z-component.
Because we have an infinite solenoid, every piece of current above r has a corresponding piece below, so all x
and y components will cancel, and the total magnetic field will point in the z direction. Since we never assumed
that r was either inside or outside the solenoid, this result holds in both cases. Finally, we can use Ampere’s
law just like in example 5.9 to conclude B = 0 outside the solenoid and B = µ0 nI ẑ inside.

loop 1
I

~r
1

r
y

~r
2

loop 2
x
I

Problem 6. Griffiths 5.17

For the toroid, N/2πs ≈ n as long as the radius of the whole toroid is very large compared to the “radius” of
the cross-sectional area. This means that s is about the same at the inner and outer edges of the toroid; in
other words, that the coils are not much closer to each other on the inside edge than on the outer edge. If this
is the case, then equation (5.58) gives B = µ0 nI just like for a straight solenoid.

2
7. Griffiths 5.20 Ampere’s law says ∇× B = µ0 J. Taking the divergence of both sides we get ∇· (∇× B) =
µ0 ∇· J = −µ0 ∂ρ
∂t , after using the continuity equation. This is inconsistent with the fact that the divergence of
a curl is always zero, unless we have ∂ρ ∂t = 0, which means we are in the magnetostatic regime. Thus outside
of magnetostatics we need to have something else on the right hand side for Ampere’s law to be valid; later
we’ll find out we have to add µ0 0 ∂E∂t . The other Maxwell equations are fine: ∇× E = 0 ⇒ ∇· (∇× E) = 0 is
consistent, and there aren’t any vanishing second derivatives we can make acting on a divergence.

8. Griffiths 5.21 At this stage we’ve just learned about electro- and magnetostatics, so we can consider Maxwell’s
equations without the time derivatives. Gauss’s Law and Ampere’s Law would probably stay the same. In
analogy with Gauss’s law, the divergence of B would be given by magnetic charges, ρm . Let the constant be
q q
α0 . Then ∇· B = α0 ρm . This leads to an analog of Coulomb’s law, F = α 0 m1 m2
4π r2 r̂. So by defining a unit
of magnetic charge we could measure the force between unit charges at a given distance in order to determine
α0 . The moving magnetic charges would presumably create electric fields, in analogy with Ampere’s law, so
∇× E = β0 Jm , where β0 is the constant we would have to measure and Jm is the magnetic current density. We
could determine β0 by measuring the force between two wires carrying a specified amount of magnetic current.
If magnetic charge is conserved, the there should be a corresponding continuity equation: ∇· Jm = − ∂ρ∂tm .
To get the force law, the first guess for the force on a magnetic charge qm could be qm [B + (v × E)]. However,
the dimensions are wrong, because E has the same units as vB. So we need to divide the second part, (v × E)
by something with dimensions of velocity-squared. The obvious choice is the speed of light, especially in light
of the relationship µ0 0 c2 = 1. So the total force law would be:
 
1
F = qe [E + (v × B)] + qm B − 2 (v × E) .
c

(The minus sign is to keep consistent with special relativity. For more discussion of magnetic charge in terms
of the full Maxwell equations, you could look ahead to Section 7.3.4 in the text.)

3
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink)
Problem Set 6

1. Griffiths 5.24

2. Griffiths 5.25

3. Griffiths 5.39

4. Griffiths 5.41

5. Griffiths 5.56

6. Griffiths 6.10

7. Griffiths 6.12

8. Griffiths 6.13
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink )

Solution Set 6 (compiled by Daniel Larson)

1. Griffiths 5.24 If B is uniform, then it is not a function of position, so any derivative of it vanish. In particular,
∇× B = 0 and ∇· B = 0. One can also check using cartesian coordinates that ∇× r = 0 and ∇· r = 3. Using
these results we find ∇· A = − 12 ∇· (r × B) = − 12 [B · (∇× r) − r · (∇× B)] = 0. Also, using the fact that
1 1

A = − 2 ∇× (r × B)= − 2 [(B · ∇)r − (r · ∇)B + r(∇· B) − B(∇· r)] =
(r · ∇)B = 0 since B is uniform, ∇×
− 12 [(B·)r − 3B]. Now, (B · ∇)r = Bx ∂x
∂ ∂
+ By ∂y ∂
+ Bz ∂z (x x̂ + y ŷ + z ẑ) = Bx x̂ + By ŷ + Bz ẑ = B. Thus
∇× A = − 21 (B − 3B) = B. We can add any constant to bf a without changing the divergence and curl, so the
result is unique up to the addition of a constant vector field.

2. Griffiths 5.25

(a) Let’s assume that A points in the same direction as the current, namely the ẑ direction. Furthermore, the
vector potential should be independent of φ and z because the infinite wire is symmetric with respect to
translations and rotations about the z-axis. So we make the guess that A = A(s) ẑ. Using the formulas for
taking divergence and curl in cylindrical coordinates, we find ∇·A = ∂z ∂
A(s) = 0 and ∇×A = − ∂s ∂
A(s) φ̂.
Since B = 2πs φ̂, we must have ∂s = − 2πs ⇒ A(s) = − 2π ln s. For the units to make sense, we need an
µ0 I ∂A µ0 I µ0 I

arbitrary length in the logarithm, so finally A = − µ2π 0I

ln(s/a) ẑ. (Note that putting “a” in the log is the
same as adding a constant, so it doesn’t change the divergence or curl of A.)

(b) First we need to find the magnetic field inside the wire, for s < R. Ampere’s law gives B · dl =
πs2
2 ⇒ B = 2πR2 φ̂. We assume that A is of the same form as in part a, so
µ0 Is
2πsB(s) = µ0 Ienc = µ0 I πR
2 2
∂s = − 2πR2 ⇒ A = − 4πR2 (s − b ) ẑ where b is the constant of integration. For s > R the B-field and
∂A µ0 Is µ0 I

thus A look the same as in part (a), exceptthat we need A to be continuous at s = R. We can accomplish
2 2
− 4πR
µ0 I
2 (s − R ) ẑ, for s ≤ R;
this by taking a = b = R. So finally, A =
− 2π ln(s/R) ẑ,
µ0 I
for s ≥ R.

3. Griffiths 5.39

(a) Using the right-hand-rule, positive charges will be deflected down.

(b) Charge accumulates on the bottom and top plates until the electric force balances the magnetic force.
For a single charge, this means qE = qvB ⇒ E = vB. The field between two large, charged plates is
essentially uniform, hence V = Et. So V = vBt. The bottom is at a higher potential, because that is
where the positive charge is.
(c) A current flowing to the right can be considered as positive charges flowing right or negative charges
flowing left. If negative charges flow left, the will also feel a magnetic force downward, and thus negative
charges will build up on the bottom plate. The potential difference between the top and bottom will be
the same, but this time the top plate will be at higher potential.

4. Griffiths 5.41 In cylindrical coordinates B is in the ẑ direction (either into or out of the page) and depends
only on the radial distance s. The particle traveling in the shaded region is assumed to be in the x − y plane
at a location specified by the coordinate r, with tangent vector dl = dr r̂ + rdφ φ̂. If the particle starts from
the origin, it cannot have any angular momentum relative to the origin. If it emerges from the shaded region
on a radial trajectory, its angular momentum is r × p = 0. So if we can show that the particle acquires no
angular momentum throughout its motion, we will have proven that it must emerge on a radial trajectory. We
 
also know that B · da = B2πr dr = 0. Recall that the torque about the origin is N = dL dt = r × F.
      
dL
L= dt = (r × F) dt = r × q(v × B) dt = q r × (dl × B) = q (r · B) dl − B(r · dl) ,
dt

1
 where we have used vdt = dl and the BAC-CAB rule for a triple cross product. Now, since the particle
is in the xy-plane and B is normal to the page, r · B = 0. Also, r · dl = r r̂ · (dr r̂ + rdφ φ̂) = r dr. So
 
L = − 2πq
B2πr dr = 0 because Bx = By = 0 and Bz 2πr dr = 0 by assumption. Thus the particle emerges
with zero total angular momentum, which means it must be traveling along a radial line.

5. Griffiths 5.56

(a) The angular momentum of a ring is L = Iω ẑ with I = M R2 , and its dipole moment will be m = IA ẑ =
2 1 2
2π/ω πR ẑ = 2 QωR ẑ. Thus m = 2M L. So the gyromagnetic ration is g = 2M .
Q Q Q

(b) Because g is independent of the radius, the same applies to all infinitesimal rings of charge. We could
calculate the total angular momentum of a spinning sphere by adding up the contributions from each ring,
just as we could get the total magnetic moment by adding up the contributions from each ring. Since each
ring will contribute to the magnetic moment and angular momentum in the same proportion, the ratio of
Q
total dipole moment to angular momentum will be the same as in part (a), g = 2M .
(c) If the electron has angular momentum 12 h̄ then the dipole moment m will be

e 1 eh̄ (1.60 × 10−19 C)(1.05 × 10−34 Js)

m= h̄ = = = 4.61 × 10−24 A m2 .
2me 2 4me 4(9.11 × 10−31 kg)

6. Griffiths 6.10 Because the magnetization is uniform, ∇× M = 0, so there is no volume bound current, but
only a surface bound current Kb = M , wrapping around the rod like the current in a solenoid. For a  L, a
is much smaller than the radius of the toroid, so in equation (5.58), we can treat s as the radius of the toroid.
NI
Then 2πs is the amount of current flowing around the toroid, per unit length, which is exactly what we mean
by surface current. Thus the B-field inside a complete, magnetized toroid is B = µ0 2πsNI
φ̂ = µ0 Kb φ̂ = µ0 M.
But part of the toroid is cut out, which we can treat as a bunch of square loops carrying the opposite current;
hence they will produce a magnetic field in a direction opposite to the one produced by the rest of the toroid.
√
In problem (5.8) we found the B-field at the center of a square loop: B = µ0 I 2/πR. In this case R = a/2
(the perpendicular distance from the center of the loop to its side). We assume that w  a, so we can think of
the gap as a single square loop with all the current running around it. Thus I = Kb w
= M w.√ Sothe missing
√
piece of the toroid contributes −2 2µ0 M w/πa. So at the center of the gap, B = µ0 M 1 − 2 πa2w .

7. Griffiths 6.12

(a) There is a surface bound current Kb = M × n̂ = kR φ̂ and a volume current Jb = ∇ × M = −k φ̂.

Since all the current is circumferential, we can think of the situation as the superposition of lots of coaxial
solenoids of different radii. So immediately we conclude B = 0 outside the cylinder. Now we can draw
a square amperian loop that has one side parallel to the z-axis inside the cylinder, and the opposite side
parallel to the z-axis outside. We know the B-field should be pointing in the z-direction, so we’ll get no
contribution to the line integral from the other two sides. Since B = 0 outside, the only section of the

loop that contributes is the piece inside the cylinder parallel to the z-axis. B · dl = BL = µ0 Ienc =


µ0 Jb da + Kb L = µ0 [−kL(R − s) + kRL] = µ0 kLs. (L(R − s) is the area of the amperian loop inside
the cylinder.) So B = µ0 ks ẑ inside.

s L

Problem 7. Griffiths 6.12

2
 (b) Since M is the only object in this problem that picks out a direction in space, we know H must also point

in the z-direction. However, using the same amperian loop as in part (a), H · dl = HL = µ0 Ifenc = 0
because there are no free currents. Thus H = 0, so B = µ0 M. Outside, M = 0 so B = 0; inside M = ks ẑ,
so B = µ0 ks ẑ.

8. Griffiths 6.13 We assume that the cavities as small enough so that the fields are essentially uniform inside of
them. We treat the cavities by considering the superposition of a piece of material without cavities and small,
cavity-shaped objects with opposite magnetization.

(a) The B-field of a uniformly magnetized sphere is 23 µ0 M, so the contribution to the B-field from the cavity
is the same as the contribution from a uniformly magnetized sphere with magnetization −M, namely
Bcav = − 32 µ0 M. Thus with the sphere removed B = B0 − 23 µ0 M. Inside the real cavity, H = µ10 B
because there is no magnetization, so = µ10 (B0 − 23 µ0 M) = H0 + M − 23 M ⇒ H = H0 + 13 M.
(b) For a long, thin, cylindrical cavity with uniform magnetization −M there is only surface current Kb = −M ,
which looks like a solenoid. So the B-field at the center is µ0 Kb = −µ0 M . Adding this to the contribution
from the cavity-less material, we find B = B0 − µ0 M. Then H = µ10 B = µ10 (B0 − µ0 M) = µ10 B0 − M ⇒
H = H0 .
(c) For the wafer shaped cavity, the bound currents run around the outside edge, so if the wafer has a large
radius and is very thin, those currents will be very small and far away from the center and will contribute
virtually no magnetic field. Thus B = B0 . Then H = µ10 B0 = H0 + M ⇒ H = H0 + M.

3
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink)
Problem Set 7

1. Griffiths 6.17

2. Griffiths 6.21

3. Griffiths 6.26

4. Griffiths 7.3

5. Griffiths 7.7

6. Griffiths 7.11

7. Griffiths 7.17

8. Griffiths 7.48
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink )

Solution Set 7 (compiled by Daniel Larson)

1. Griffiths 6.17 In a linear material, we know H is proportional to B: B = µH = µ0 (1 + χm )H, so for a long

wire it should be circumferential. We can then use Ampere’s law to find H from the free current, and then get
B from H. As usual, we draw an amperian loop around the wire:


I(s2 /a2 ), (s < a)
H · dl = 2πsH(s) = Ifenc =
I (s > a)
 
Is µ0 (1+χm )Is
H(s) = 2πa2 , (s < a)
⇒ B(s) = 2πa2 , (s < a)
.
I µ0 I
2πs , (s > a) 2πs , (s > a)

In a linear material, Jb = χm Jf = χm πaI 2 (using the fact that I is uniform over the area of the wire) and points
in the same direction as I. Kb = M × n̂ = χm H(a) × n̂ ⇒ Kb = χ2πa mI
in the direction opposite from I (using
2
the right-hand-rule). The total bound current is Ib = πa Jb + 2πaKb = χm I − χm I = 0 as it must be.

2. Griffiths 6.21

(a) We need to compute the work it takes to bring the magnetic dipole in from infinity to the origin and
rotate it to its final configuration. First, bring the dipole to the origin along a trajectory in which m is
always perpendicular to B so that there is no force on the dipole and hence no work done. For simplicity,
imagine B is uniform and points in the ŷ direction. Then we can slide a dipole (pointing in the x̂
direction) in along the x-axis. All the work comes from rotating the dipole in the presence of the B-field.
The torque exerted by the B-field is N = m × B = mB sin θ ẑ where θ is the angle between m and B
(initially π/2); this is opposite the torque we must exert in order to rotate the dipole. So to move the
dipole from an angle of π/2 with respect to B to some other angle θ we must do an amount of work
θ
U = π/2 mB sin θ
dθ
= mB(− cos θ
)|θπ/2 = −mB cos θ = −m · B.
(b) We can put the first diple at the origin. It produces a magnetic field B1 = 4πr µ0
3 [3(m1 · r̂) r̂ − m1 ] at

any location r. The second dipole, located at r, interacts with this magnetic field as in part (a). Thus
U = −m2 · B1 = − 4πr
µ0
3 [3(m1 · r̂)m2 · r̂ − m2 · m1 ] = 4πr 3 [m1 · m2 − 3(m1 · r̂)(m2 · r̂)].
µ0

(c) From the figure, mi · r̂ = mi cos θi for i = 1 or 2, and m1 · m2 = m1 m2 cos(θ1 − θ2 ) = m1 m2 (cos θ1 cos θ2 +
sin θ1 sin θ2 ). So U = µ04πr
m1 m2
3 [cos(θ1 − θ2 ) − 3 cos θ1 cos θ2 ] = µ04πr
m1 m2
3 [sin θ1 sin θ2 − 2 cos θ1 cos θ2 ]. A sta-
ble configuration occurs when the energy is at a minimum.
∂U µ0 m1 m2
= (cos θ1 sin θ2 + 2 sin θ1 cos θ2 ) = 0 ⇒ 2 sin θ1 cos θ2 = − cos θ1 sin θ2
∂θ1 4πr3
∂U µ0 m1 m2
= (sin θ1 cos θ2 + 2 cos θ1 sin θ2 ) = 0 ⇒ 2 sin θ1 cos θ2 = −4 cos θ1 sin θ2
∂θ2 4πr3
So we need cos θ1 sin θ2 = sin θ1 cos θ2 = 0. This will happen for either sin θ1 = sin θ2 = 0 ⇒ (i) →→ or
(ii) →←; or if cos θ1 = cos θ2 = 0 ⇒ (iii) ↑↑ or (iv) ↑↓. We know that the lowest energy configuration
will have m lined up with B. This only happens in (i) and (iv), so they are the stable minima. To find
the absolute minimum, we need to calculate U . For situation (i) we have θ1 = θ2 = 0 so U = µ04πrm1 m2
3 (−2)
whereas for (iv) we have θ1 = −θ2 = π/2, so U = 4πr3 (−1). Thus the most stable configuration is the
µ0 m1 m2

one with the lowest energy, namely (i) where the magnetic moments are lined up along the line joining
them: →→.
(d) Using the result from part (c), the most stable configuration should be when the dipoles all form one line,
pointing in one direction: →→→→→.

1
3. Griffiths 6.26 The angle θ1 is related to the components of B1 which are parallel and perpendicular to the


B1
interface: tan θ1 = B1⊥
. The same relation holds for θ2 and B2 . The perpendicular components of B are
continuous across the boundary, so B1⊥ = B2⊥ . We also know that the parallel components of H are continuous
  
across the boundary, since there is no free surface current. Since B = µH this gives: H1 = H2 ⇒ µ11 B1 =
1 
mu2 B2 . Putting these together:
 
tan θ2 B2 B1⊥ B µ2
= ⊥ 
= 2 =
tan θ1 B2 B1 B1 µ1

4. Griffiths 7.3

(a) To find the resistance, we need to look at the ration of the potential difference to the current flowing
between to metal objects. Any currents flowing will leave conductor 1 and flow to conductor 2. So we can

find the current by enclosing conductor 1 with a surface and then evaluating I = J · da. This equation

is exactly what we need. First, Gauss’s law tells us E · da = 10 Qenc , while Ohm’s law gives J = σE
and V = IR. We assume there are no free charges floating around in our conducting material, so Qenc
is simply the charge on the first object, which is related to the capacitance of the system by Q = CV .
These are all the ingredients we need.
 
σ σ σ 0
I = J · da = σ E · da = Q = CV = CIR ⇒ R = .
0 0 0 σC

(b) We apply a potential difference V0 between objects 1 and 2 and then allow the charge to leak off. The
voltage at any time is given by V (t) = I(t)R = − dQ dt R, where the minus sign comes because we assume
the current I is positive, but we know the charge Q is decreasing. We also know that V = Q/C, so that
1 dQ 1
dt = C dt , because capacitance is just a constant. Thus V (t) = −RC dt ⇒ dt = − RC V (t) ⇒
tells us dV dV dV

V (t) = V (0)e−t/RC = V0 e−t/RC . Then the time constant τ = RC = 0 /σ.

5. Griffiths 7.7

(a) Current will flow due to the changing flux in the loop formed by the bar and the wire. The total flux
through the loop is Φ = BA. If the bar is moving at speed v to the right, the area is changing at a rate
dt = lv. Thus E = − dt = −Blv. Then E = IR ⇒ I = Blv/R. The minus sign just refers to the
of dA dΦ

direction, but it is easier to figure that out using Lenz’s law. Since the flux into the page is increasing,
the current will flow to produce flux coming out of the page, so the current will be going down through
the resistor.
(b) There is magnetic force on the bar because there is a current flowing in the presence of a magnetic field.

F = Idl × B = IlB = B 2 l2 v/R and it points to the left, which is the direction of dl × B.
(c) The force on the bar is slowing it down so we take it to be negative.

1 22 dv dv B 2 l2 2 2
F =− B l v = ma = m ⇒ =− v ⇒ v(t) = v0 e−B l t/Rm .
R dt dt Rm
(d) The energy goes into heading the resisitor. The power delivered to the resisitor is

dW B 2 l2 2 −2αt B 2 l2 dW
P = = I 2R = v0 e , where α = ; ⇒ = αmv02 e−2αt .
dt R Rm dt
The bar keeps slowing down, but takes an infinite amount of time to stop. During this time, the total
energy delivered to the resistor is
 ∞ 
−2αt ∞
2 −2αt 2 e  = αmv02 1 = 1 mv02 .
W = αmv0 e dt = αmv0
0 −2α 0 2α 2

2
6. Griffiths 7.11 Let l be the width of the loop, and s be the distance between the top edge of the loop and
the bottom of the region of B-field. The flux through the loop is Φ = Bla, so E = − dΦ dt = −Bl dt . Let’s only
ds

consider magnitudes and drop the minus sign. Since ds dt = v(t), the velocity of the loop at time t, we have
E = Blv = IR, assuming the loop has resistance R. Then I = Blv/R is the current flowing in the loop. As
the loop falls, the flux into the page is decreasing, so the current flows in a clockwise direction to oppose the
change in flux. But the part of the loop still in the region of magnetic field will feel a force because there is
a current in a magnetic field. The forces on the two sides will cancel, leaving an upward force of magnitude
F = IlB = B 2 l2 v/R. This force opposes the force of gravity, Fg = mg which pulls the loop downward. The loop
will have reached terminal velocity, vt , when these two forces balance: mg = B 2 l2 vt /R ⇒ vt = (mgR)/(B 2 l2 ).
2 2
To find the velocity as a function of time, we need Newton’s second law: Fnet = ma = m dv dt = mg − R v
B l

where I have taken the downward direction to be positive. Letting α = B 2 l2 /mR, we have vt = g/α, and we
get a differential equation for the velocity:
dv dv 1
= g − αv ⇒ = dt ⇒ − ln(g − αv) = t + const. ⇒ g − αv = Ae−αt
dt g − αv α

Since the loop starts at rest at t = 0, the constant A = g. Thus v(t) = αg (1 − e−αt ) = vt (1 − e−αt ). At 90% of
terminal velocity we have v/vt = 0.9 = 1 − e−αt ⇒ e−αt = 0.1 ⇒ t = α1 ln 10 = vgt ln 10.
To get a numerical answer, we need various properties of aluminum and the dimensions of the loop. Assume
the loop is square, with sides l and cross-sectional area A. The resistivity is ρ = σ1 = 2.65 × 10−8 Ω m; the
mass density is η = 2.7 × 103 kg/m3 ; g = 9.8 m/s2 ; and B = 1 T. The resistance of a piece of metal with
uniform cross-sectional area A and length L is R = Aσ L
, so in this case we have R = 4lρ
A .

mgR (ηA4l)g(4lρ/A) 16ηgρ vt

vt = 2 2
= 2 2
= 2
= 1.1 cm/s; ⇒ t90% = ln 10 = 2.8 ms
B l B l B g
Finally, if the loop were cut, no current would flow, so there wouldn’t be any force to oppose gravity and the
loop would fall freely under the force of gravity.

7. Griffiths 7.17

(a) We assume that the solenoid is relatively long, so the only magnetic field in the loop is the uniform
B-field inside the solenoid, namely B = µ0 nI. Thus the flux passing through the loop is Φ = πa2 B =
πa2 µ0 nI ⇒ E = −πa2 µ0 n dI dt . The negative sign just refers to the direction, which is easier to find
using Lenz’s law, so we’ll ignore it. The magnitude of the current passing through the resistor is given by
E = Ir R ⇒ Ir = R1 πa2 µ0 nk. The flux due to the solenoid is pointing to the right and is increasing, thus
the current in the loop will flow in order to produce a flux inside the loop pointing to the left, which is
opposite the direction of the current flowing in the solenoid, or to the right in the picture in the text.
(b) When the solenoid is pulled out and reinserted there is lots of changes going on in the flux, most of them
very complicated. But all we need to know to get the total charge is the total change in flux.
  
E 1 dΦ 1 1
∆Q = I dt = = − = − (Φf − Φi ) ⇒ ∆Q = ∆Φ (in magnitude)
R R dt R R

Initially there is flux Φi = πa2 µ0 nI pointing to the right, and at the end there is the same amount of
flux pointing in the opposite direction, the net change in flux is ∆Φ = 2πa2 µ0 nI, which means ∆Q =
1 2
R 2πa µ0 nI.

8. Griffiths 7.48 Starting with Equation (5.3), we have qBR = mv. Keeping R fixed, we can differentiate
with respect to time: qR dB dv
dt = m dt = ma = F = qE. Thus E = R dB
dt , where B is evaluated at the radius

of the electron’s orbit, R. From Faraday’s law we know E · dl = − dΦdt , so if we take the loop to be the
electron’s orbit at radius R, 2πRE = − dt . Combining this with the previous result we can solve for B:
dΦ

3
 1 dΦ
 Φ 
dB
= − 2πR 2 dt ⇒ B = − 12 πR 2 + C where C is some integration constant. If B = 0 when t = 0, there
dt  Φ 
will be no flux through the loop, so the constant must be zero. But this means B(R) = − 12 πR 2 . The term

in parentheses is simply the total field throughout the orbit (flux) divided by the area of the orbit, namely
the average field. Thus the average field over the orbit is twice the value of the field at the circumference (in
magnitude).

4
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink)
Problem Set 8

1. Griffiths 7.21

2. Griffiths 7.23

3. Griffiths 7.26

4. Griffiths 7.58

5. Griffiths 7.30

6. Griffiths 7.31

7. Griffiths 7.37

8. Griffiths 7.50
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink )

Solution Set 8 (compiled by Daniel Larson)

1. Griffiths 7.21 The emf is the time-derivative of the flux due to the small loop that passes through the big
loop, namely E = − dΦdt = −M dt = −M k where we have used the definition of the mutual inductance, that the
dI

flux through the big loop is proportional to the current in the small loop, and the proportionality constant is
M . So all we need to do is calculate M . However, it is quite difficult to calculate the flux due to the small
loop, because the B-field due to a square loop is rather complicated. Instead, we can use the equality of mutual
inductances and find M by calculating the flux through the small loop due to a current in the big loop. This is
much easier, because the big loop is essentially two long wires, symmetrically placed on either side of the little

2a 1
loop. The field from one long wire is B = µ2πs
0I
⇒ Φ1 = µ2π 0I
a s
a ds = µ2π
0 Ia
ln 2. The flux from the two wires
are the same, so we multiply the above result by 2 and divide by I to find M = (µ0 a ln 2)/π). The magnitude
of the emf is then E = (µ0 ka ln 2)/π.
To determine the direction of the current in the big loop we will use Lenz’s law, and also the fact that the
B-field lines produced by the square loop are all closed curves. That means that every line of flux heads into
the page inside the square loop and comes back out of the page somewhere outside the square loop. Since
the big loop encloses all of the flux heading into the page through the center of the small square loop, but
only some of the flux coming back out of the page, the net flux through the big loop is into the page. When
the current in the small loop decreases, the net flux into the page decreases, so a counterclockwise current is
induced in the large loop to oppose the change in flux.

2. Griffiths 7.23 We need to compute the flux passing through the loop due to the current flowing in the two long
sides. The field from a single long wire is B = µ2πs
0I
, so we need to integrate this from 0 to d and then multiply by
2 because the top and bottom wire both contribute to the flux in the same direction. The flux from one wire is

d 1 d

thus Φ1 = µ2π
0I
0 s
l ds = µ0 Il
2π ln s . However, when we try to evaluate the natural log at 0 we find it diverges.
0
So we introduce a small thickness  to the wires, and integrate from  to d − . Thus Φ1 = µ2π 0 Il
ln d−
. We can
ignore the  in the numerator, because it is tiny compared to d, but it is very important in the denominator.
Multiplying Φ1 by 2 (for the two wires) and dividing by I gives us L = µπ0 l ln(d/). The size of the wire is very
important in determining L!

3. Griffiths 7.26

(a) The field inside a solenoid is B = µ0 nI, so the flux through a single turn of the wires is Φ1 = πR2 B =
πR2 µ0 nI. The total flux through a section of length l is the flux through one turn times the total number
of turns, N = nl. Thus Φ = N Φ1 = πR2 µ0 n2 lI ⇒ L = πR2 µ0 n2 l. Finally, W = 12 LI 2 = 12 πR2 µ0 n2 lI 2 .

(b) W = 12 (A · I) dl where A(R) = (µ0 nI/2)R φ̂. We need to do the integral over the whole “loop”, which
we can do by integrating around a single turn and multiplying by the total number of turns, N = nl. So

for a single turn, W1 = 14 µ0 nIR φ̂ · I φ̂R dφ = 14 µ0 nIR2πRI = 12 πµ0 nI 2 R2 . Multiplying by N gives
W = 12 πR2 µ0 n2 lI 2 .



(c) W = 2µ1 0 all space B 2 dτ = 2µ1 0 solenoid (µ0 nI)2 dτ = 2µ1 0 µ20 n2 I 2 πR2 l = 12 πR2 µ0 n2 lI 2 .

 
(d) W = 2µ1 0 V B 2 dτ − S (A × B) · da . The volume is a cylindrical tube from radius a < R to b >
R. The only B-field is inside the cylinder, so the calculation of the first term is identical to part (c)
but instead of the whole volume πR2 l we have only the volume for s > a, namely π(R2 − a2 )l. So

2
B dτ = µ20 n2 I 2 π(R2 − a2 )l. Now for the second term. Since B = 0 outside the solenoid, we only
need to worry about the inner surface at s = a. A(a) = 12 µ0 nIa φ̂ and B = µ0 nI ẑ, so A × B =


2 µ0 n I a( φ̂ × ẑ) = 2 µ0 n I a ŝ. Thus (A × B) ·  da = ( 12 µ20 n2 I 2 a ŝ) · [adφ dz(− ŝ)] = − 12 µ20 n2 I 2 a2 2πl.
1 2 2 2 1 2 2 2

Finally, W = 2µ1 0 µ20 n2 I 2 π(R2 − a2 )l + µ20 n2 I 2 a2 πl = 12 πR2 µ0 n2 lI 2 . All four methods agree!

1
4. Griffiths 7.58

(a) The ribbon looks like a long, parallel plate capacitor. If there is surface charge +σ on the top, and −σ
on the bottom, the field between the “plates” is E = σ/0 , which means the potential difference between
them is V = Eh = σh/0 . In a length L, the charge per area is σ = Q/wL. Thus C = Q/V = Qh/ Q0 wL ⇒
C = C/L = 0 w/h.
(b) If there is uniform surface charge K flowing down the top ribbon and back up the bottom, they produce a
B-field between the ribbons which is approximately uniform and points perpendicular to the current. Over
a small amperian loop, BL = µ0 KL ⇒ B = µ0 K = µ0 I/w. The flux passing between the ribbons, in a
length l (now measured along the length of the ribbons) is Φ = Bhl = µw0 I hl = LI ⇒ L = L/l = µ0 h/w.
running width-wise
(c) LC = µ0 h 0 w
w h = µ0 0 = 1/c2 = 1.11 × 10−17 s2 /m2 . The speed of propagation is c.
(d) When a dielectric is present, the capacitance is multiplied by the dielectric constant, r : C  = r C =
r 0 w/h = w/h. The inductance works the same way. With the magnetic material present, H = K, so
B = µH = µK instead of µ0 K. So we just replace 0 and µ0 with  and µ. LC = µ. The propagation
√
speed is v = 1/ µ.

5. Griffiths 7.30

(a) Treating the small loops as magnetic dipoles, the magnetic field due to loop 1 is B1 = 4πr µ0
3 [3(m1 · r̂) r̂ − m1 ] =

4πr 3 [3((a1 · r̂) r̂ − a1 ], where r is the vector from loop 1 to loop 2 and m1 = I1 a1 . If loop two is
µ0 I1

very small, then the magnetic field is essentially constant over its area, and we have Φ2 = B1 · a2 =
4πr 3 [3(a1 · r̂)(a2 · r̂) − a1 · a2 ] = M I1 . Thus M = 4πr 3 [3(a1 · r̂)(a2 · r̂) − a1 · a2 ]
µ0 I1 µ0

(b) We want to keep a constant current I1 in loop 1. However, turning on a current in loop two causes an
emf in loop 1: E1 = −M dI 2
dt . The induced emf does work at a rate P = dW dt = I1 E1 , so the work done per

dW 
unit time against the induced emf is opposite this, dt 1 = −I1 E1 = M I1 dI 2
dt . Since I1 is assumed to be
constant, we can integrate this equation to get the total work done. Since the current in loop 2 starts at
3 [3(m1 · r̂)(m2 · r̂) − m1 · m2 ]. This
µ0
zero and increases to a final value of I2 , we have W1 = M I1 I2 = 4πr
is the total energy of interaction between the two loops, which is opposite in sign to the result in equation
6.35 in the text. The reason is that there we derived the interaction energy of two fixed dipoles. The
only work we needed to do to assemble the system was move one dipole around in the field of the other
dipole. But in the current problem, we also included the work necessary to maintain the dipole moment
of one loop in the presence of the other. It is a funny coincidence that the only difference between the
two calculations is a minus sign.

6. Griffiths 7.31 The displacement current density is Jd = 0 ∂E ∂t . If there is a surface charge density +σ on the
one side of the gap, and −σ on the other side, the electric inside the gap is approximately like that of a parallel
∂
plate capacitor, namely uniform inside, zero outside, and with magnitude σ/0 . Then Jd = 0 ∂t E = ∂σ
∂t =
I/A = I/(πa ), pointing in the ẑ direction. Then if we draw an amperian loop with radius s in the gap, there
2

is no conduction current flowing through the loop, and we can use the extended Ampere’s law:

I s2 µ0 Is2 µ0 Is
B · dl = B(s)2πs = µ0 Idenc = µ0 2 πs2 = µ0 I 2 ⇒ B = 2
⇒ B= φ̂
πa a 2πsa 2πa2

7. Griffiths 7.37 We place a parallel plate capacitor in the sea water and connect it to a voltage source. There
will be some normal conduction current, Jc , due to electrons in the sea water traveling from one plate to the
other; but there will also be some displacement current, Jd , due to the changing electric fields. First we find
the conduction current. If the potential difference between the plates is V then the electric field is E = V /d

2
 where d is the distance between the plates. Then Jc = σE = ρ1 E = V0 cos(2πνt)
ρd . On the other hand, treating
the seawater as a linear dielectric, the displacement current is given by
 
∂D ∂ ∂ V0 cos(2πνt) V0
Jd = = (E) =  = [−2πν sin(2πνt)].
∂t ∂t ∂t d d

We are only interested in determining which contribution is bigger, so we don’t need to worry about the time
dependence or the fact that the currents are out of phase; we just make the ration of their amplitudes:
Jc V0 d 1  −1
= = = 2π(4 × 108 )(81)(8.85 × 1012 )(0.23) = 2.41
Jd ρd V0 2πν 2πνρ
It is a good exercise to check that the units all cancel to leave a pure number.

8. Griffiths 7.50 Since we are assuming that the voltmeters draw negilgible current, there is a single square
circuit containing the two resistors surrounding the solenoid. The flux through the loop formed by the circuit
is the flux inside the solenoid, Φ = αt. Thus the emf in the loop is E = − dΦdt = −α, which drives a current
I = |E|/R = α/(R1 + R2 ) in the counter-clockwise direction. Meter 1 measures the voltage drop across R1 ,
namely V1 = IR1 = αR1 /(R1 + R2 ) (it is positive because Vb is the higher potential) and meter 2 reads
V2 = −IR2 = −αR2 /(R1 + R2 ) (Vb is at a lower potential).

3
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink)
Problem Set 9

1. Griffiths 10.3 are equivalent to the familiar formulæ

Z2 − Z1
R̃ =
2. Griffiths 10.5 Z2 + Z1
2Z2
T̃ = , where
Z2 + Z1
3. Griffiths 10.7
Ẽ0
Z≡ ,
H̃0
4. Griffiths 8.1 Ẽ0R
R̃ ≡ , and
Ẽ0I
5. Griffiths 9.11 Ẽ0T
T̃ ≡ ,
Ẽ0I

6. Griffiths 9.13 and where Z is the characteristic impedance of

the medium, R̃ is the amplitude reflection co-
efficient, and T̃ is the amplitude transmission
7. At the rate of 1 card/sec, psychic Uri Geller coefficient.
(http://skepdic.com/geller.html) turns over
each card in a deck. He communicates by “para-
normal” means the identity of each card to his
assistant, from whom he is shielded with respect
to sound and visible light.

As a physicist, you consider all EM waves to be

normal. To test the notion that Uri’s talents defy
the laws of physics, you resolve to design a shield
that will prevent Uri from using any relevant EM
frequency to communicate with his assistant.
(a) Roughly what minimum EM frequency must
Uri use? (Hint: Consider that a 56 kbps modem
operates over audio telephone frequencies.)
(b) Design a spherical shell, enclosing a volume
of 1 m3 for Uri’s comfort, that will attenuate
the EM waves generated by Uri’s brain to ≈
1/400 ≈ e−6 of their original amplitude. Use the
minimum EM frequency that you calculated in
(a).
(c) How much does your shield weigh? (Try to
design the lightest shield that will do the job.
Does it help to use a ferromagnetic material?)

8. Show that the results in Griffiths Eq. (9.147)

 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink )

Solution Set 9 (compiled by Daniel Larson)

qt
1. Griffiths 10.3 First we’ll calculate the fields. E = −∇V − ∂A 1 1 r̂
∂t = 0 − 4π0 r 2 r̂. B = ∇× A = − 4π0 ∇× r 2 = 0.
These fields should be familiar, because they are the fields of a stationary point charge q at the origin. Thus
ρ = qδ 3 (r) and J = 0. If you took the divergence of E to find ρ you got zero, which is correct everywhere
except at the origin, where E blows up and derivatives are ill defined. The delta-function at the origin is the
function that has zero divergence everywhere except the origin, which is why we use it to represent the charge
density for a point charge.

2. Griffiths 10.5 We’re given λ, so we can calculate the new potentials.




1 qt 1 1 1 qt 1 1
A = A + ∇λ = − r̂ − qt∇ =− r̂ + qt 2 r̂ = 0.
4π0 r2 4π0 r 4π0 r2 4π0 r


∂λ 1 q 1 q
V
=V − =0− − = .
∂t 4π0 r 4π0 r
These are the more familiar potentials for the point charge of Problem 1.

3. Griffiths 10.7 We want ∇· A = −µ0 0 ∂V ∂t , so first assume that it isn’t true and then prove we can make a
gauge transformation so that it becomes true. So assume that ∇· A + µ0 0 ∂V ∂t = Φ, where Φ is a function that
we know. For any λ, the new vector potential is A = A + ∇λ, while V = V − ∂λ



∂t , and we want to require

∂V

∇· A + µ0 0 ∂t = 0. Note that both the vector potential A and the scalar potential V transform. Thus

∂V
∂V ∂2λ
∇· A
+ µ0 0 = ∇· A + ∇2 λ + µ0 0 µ0 0 2 = Φ + ✷2 λ = 0 ⇒ ✷2 λ = −Φ.
∂t ∂− ∂t
This equation is of the form (10.16 (i)), so assuming we can find a solution λ, we can make a gauge transfor-
mation using λ to get into Lorentz gauge.
t
If we choose λ = 0 V dt
then V
= V − ∂λ ∂t = V − V = 0, so we can always find a gauge in which the scalar
potential vanishes. This doesn’t cause any problems, because the electric field gets contributions from both the
scalar potential and vector potential, so for a given E-field we can choose V = 0 and still get the proper E from
A. However, if we have a non-zero B-field, then finding a gauge where A = 0 would mean B = ∇× A = 0,
which would be a contradiction. Thus we cannot in general find a gauge in which A = 0.

4. Griffiths 8.1

(a) From Example 7.13, we have a cable consisting of two concentric cylinders. The B-field between them is
B = µ2πs
0I
φ̂. For linear charge density λ on the inner cylinder, the electric field between the cylinders is
E = 2πλ0 s ŝ. Thus
1 λI
s= (E × B) = ẑ.
µ0 4π 0 s2
2

We want to find the power being transported down the cable, so we need to integrate s over a cross-section
of the cable perpendicular to the z-axis. s is only nonzero between the cylinders.
  b  b

λI ds λI b
P = s · da = S 2πs ds = = ln .
a 2π 0 a s 2π 0 a

To express this in terms of the potential difference V , look back at the solutions to Homework #2, Problem
λ
 
2.39 where we found that V = |V (b) − V (a)| = 2π 0
ln ab . Substituting into our result, this gives P = IV .

1
 (b) We proceed exactly as in part (a). For surface charge σ on the ribbons, E = σ0 ẑ. Also, B = µ0 K x̂ = µw0 I x̂
(see the solution to problem 7.58). Thus s = µ10 (E × B) = σI 0w
ŷ. Again, the power transported is the
surface integral of s over a cross-section perpendicular to the length of the ribbons. In this case, s
is constant so we just multiply by the area, wh. Thus P = σIh/0 . But the potential difference is

V = − E · dl = σh/0 , which again gives P = IV .

5. Griffiths 9.11 We want to compute the time average of f = A cos(k · r − ωt + δa ) multiplied by g = B cos(k ·
r − ωt + δb ) over one period, T . First, lets do it the long way.
 T  T
1 AB
f g = A cos(k·r−ωt+δa ) B cos(k·r−ωt+δb )dt = [cos(2k · r − 2ωt + δa + δb ) + cos(δa − δb )] dt.
T 0 2T 0

Here I used the trig identity: cos(α) cos(β) = 12 (cos(α + β) + cos(α − β)). Now the integral isn’t too hard. The
first term is the integral of a cosine over one full period, which gives zero. The second term is independent of
t, so just gets multiplied by T . Thus f g = AB 1
2T cos(δa − δb )T = 2 AB cos(δa − δb ).

Now let’s calculate the average using complex notation. We let f = Re(f˜) where f˜ = Aei(k·r−ωt+δa ) =
Aeiδa ei(k·r−ωt) ≡ Ãei(k·r−ωt) . Similarly, g = Re(g̃) where g̃ = B̃ei(k·r−ωt) and B̃ = Beiδb . Then
1 1  i(k·r−ωt) ∗ −i(k·r−ωt)  1  iδa −iδb 
Re(f˜g̃ ∗ ) = Re Ãe B̃ e = Re Ae Be
2 2 2
AB   AB
= Re ABei(δa −δb ) = AB cos(δa − δb ) = f g.
2 2

6. Griffiths 9.13 The derivation of the exact reflection and transmission coefficients, as defined by Griffiths,
follows Section 9.3.2 in the text up through equation (9.82).



1−β 2
Ẽ0R = Ẽ0I , Ẽ0T = Ẽ0I
1+β 1+β
µ1 v1 µ1 n2
where β ≡ µ2 v2 = µ2 n1 . Griffiths defines the intensity reflection coefficient as the ratio of reflected to incident
intensity.

2
2
2
2
IR E0R 1−β µ2 v2 − µ1 v1 µ2 n1 − µ1 n2
R≡ = = = =
II E0 I 1+β µ2 v2 − µ1 v1 µ2 n1 − µ1 n2
n2 µ v n2 µ n µ1 n2
Notice that µi i c2 = n2i (Eqn 9.68). This implies 21 vv21 = µ22 n12 v21 = n22 µ12 n12 = µ2 n1 = β. Thus the intensity
1 1
transmission coefficient is
2
2
IT 2 v2 E0T 2
T ≡ = =β
II 1 v1 E0I 1+β
You can plug in β in terms of µ and n, but that isn’t too enlightening. To add the two coefficients it is easiest
to leave things in terms of β.

4β (1 − β)2 1  2
 1  2
 (1 + 2β + β 2 )
T +R= + = 4β + (1 − β) = 4β + 1 − 2β + β = =1
(1 + β)2 (1 + β)2 (1 + β)2 (1 + β)2 (1 + β)2

7. Psychic

(a) If Uri is flipping 1 card per second then he must transfer information at a rate of about 6 bits per second,
since 6 bits provides 26 = 64 different combinations, which is enough to specify a single card out of the 52
cards in a deck. Knowing that a 56 kbps modem essentially saturates the capacity of the phone network,
which carries frequencies between 200 Hz and 3000 Hz, we can assume that to carry 6 bps one would need
a bandwidth of:
6 bps
(3000 − 200) Hz = 0.3 Hz
56000 bps
To get this bandwidth, the minimum frequency Uri would need will be about 0.3 Hz ⇒ ω = 1.88 s−1 .

2
 (b) If we make a spherical shell enclosing a cubic meter, it must have an inner radius of R = (3/4π)1/3 = 0.62
m. We want to make it out of a conducting material so that the EM waves are attenuated as they travel
through the shell. From equation (9.127) in the text, we see that the amplitude is proportional to e−κz ,
where z is the direction of propagation. So to get the amplitude attenuated by 1/400 ≈ e−6 we need a
thickness t = 6/κ. From equation (9.126),

1/2
µ  σ 2
κ≡ω 1+ −1
2 ω

To make a small shell, we need κ to be large, which means choosing a material with a high conductivity.
For example, let’s use copper. (It has a high conductivity, but lower density and lower cost than silver.)
From the table section 7.1, the conductivity of copper is σ = 1/ρ = 6.0 × 107 (Ωm)−1 . Assuming  ≈ 0
and µ ≈ µ0 , we find κ = 8.4 m−1 , so the thickness we would need would be t = 6/κ = 0.71 m. That’s
quite a thick shell!
(c) Since the inner radius of the shell is 0.62 m, the outer radius would have to be 1.33 m, which gives a
 
total volume of copper V = 43 π 1.333 − 0.623 = 8.86 m3 . The density of copper is 8.96 g/cm3 = 8.96
×103 kg/m3 , which gives a total mass of (8.86)(8.96 × 103 ) = 79000 kg! That’s about 87 tons. So copper
probably isn’t the best material to use.
The suggestion to use a ferromagnetic material is a good one. Looking at the table of resistivities, we
see that iron has a conductivity σ ≈ 107 , only a factor of 6 smaller than copper. However, iron can
also have permeability µ much bigger than µ0 , which will help increas κ and thus decrease the required
shell thickness. A further bonus is that the density of iron is slightly smaller than that of copper. In
my brief research on the subject (namely typing “magnetic permeability” into Google), I found the claim
that well designed ferromagnets can have permeabilities up to µ = 106 µ0 . Taking this extreme case, and
still assuming that  ≈ 0 , we find κ = 3510 m−1 . Thus the thickness is t = 6/κ = 1.7 mm. Now the
shell requires only V = 4πR2 t = 8.2 × 103 cm3 of iron. With a density of 7.86 g/cm3 , this yields a shield
that weighs about 64 kg. Still, this isn’t light (about 142 lbs), but it is much more reasonable than the
gargantuan copper shield. Of course, we’re not likely to find that much iron with so high a permeability,
but this demonstrates how important ferromagnetic materials are for practical shielding.

8. Compare Results Griffiths has equation (9.147):


 

1 − β̃ 2 Ẽ0R 1 − β̃ Ẽ0T 2
Ẽ0R = Ẽ0I , Ẽ0T = Ẽ0I ⇒ R̃ ≡ = , and T̃ ≡ =
1 + β̃ 1 + β̃ Ẽ0I 1 + β̃ Ẽ0I 1 + β̃

Now we need to relate β̃ to Z1 and Z2 . Griffiths defines β̃ = µµ12vω1 k̃2 , while we define Z = H̃
Ẽ0
in each medium.
0
Since B = µH in linear materials, and using Griffiths’s equations (9.140)-(9.142), we find:

Ẽ0I Ẽ0 Ẽ0T Ẽ0 µ2 ω

Z1 = = 1 I = µ1 v1 and Z2 = = 1 T =
H̃0I µ1 B̃0I H̃0T B̃
µ2 0T k̃2

Thus β̃ = µµ12vω1 k̃2 = Z1 /Z2 . Plugging this into the above formulas and multiplying both numerator and denom-
inator by Z2 we get the “familiar” equations:
Z2 − Z1 2Z2
R̃ = and T̃ =
Z2 + Z1 Z2 + Z1

3
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink)
Problem Set 10

1. Griffiths 9.19

2. Griffiths 9.20

3. Griffiths 9.21

4. Griffiths 11.3

5. Griffiths 11.4. You need calculate only the

time average Poynting vector, intensity, and
total power radiated (this is much simpler than
computing the full time-dependent expressions).

6. Griffiths 11.9

7. Griffiths 11.14

8. Griffiths 11.21
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink )

Solution Set 10 (compiled by Daniel Larson)

1. Griffiths 9.19

(a) For a poor conductor, σ
ω, we can expand the second square root in the formula for κ (Eqn 9.126):

 σ 2 1  σ 2
1+ ≈1+ + ···.
ω 2 ω
Then we can evaluate κ explicitly:

 1/2


µ 1  σ 2 σ µ 1 2 
κ=ω 1+ −1 = ⇒ d= =
2 2 ω 2  κ σ µ

For pure water,  = 80.10 (Table 4.2), µ = µ0 (1 + χm ) ≈ µ0 (Table 6.1), and σ = 1/ρ = 1/(2.5 × 105 )
(Table 7.1). Plugging these numbers in, you should find d = 1.19 × 104 m.
(b) For a good conductor, σ  ω, so we can ignore the “1s” in Eqn. 9.126. In particular, notice that in this
limit κ = k, which means d = 1/κ = 1/k = 1/(2π/λ) = λ/2π. In this limit we have:

 

µ σ 1/2 ωµσ
κ≈ω = = 8 × 107 m−1 ,
2 ω 2
where I’ve used the given values. So the skin-depth is d = 1/κ = 1.3 × 10− 8 m = 13 nm.
(c) We’re still in the regime where κ ≈ k, so the phase difference is φ = tan−1 (κ/k) = tan−1 1 = 45◦ , and it
is always the magnetic field that lags behind. The ratio of their amplitudes is given by Eqn. 9.137:


2


B0  σ µσ
= µ 1 + ≈ = 10−7 s/m.
E0 µ ω

(I used the same numbers as given for part (b).) Compared to the ratio of the amplitudes in vacuum,
namely 1/c, the B-field is comparatively about 100 times larger in the good conductor.

2. Griffiths 9.20

(a) Using Eqn (9.138), and taking the time average so cos2 → 12 :

   

1 1 2 1 −2κz  σ 2
2 21 1 21 1 −2κz 2 1
u = E + B = e E0 + B0 = e E0  + µ 1 +
2 µ 2 2 µ 2 4 µ ω


 2 −2κz  σ 2  2 k2 k2
= E0 e 1+ 1+ = E02 e−2κz = E 2 e−2κz
4 ω 4 µ ω 2 2µω 2 0

(Here I used (9.126) to replace the square root with k.) We can see that the magnetic contribution
dominates by looking at the first expression in the second line, above. The 1 represents the electric

contribution, while the 1 + (σ/ω)2 comes from the magnetic contribution. Since the second term is
always greater than or equal to 1, the magnetic contribution dominates.
(b) You can do this problem by calculating S and time averaging, but I prefer to do it in a more physical way.
From (9.138) we can see that the energy is flowing in the ẑ direction, since S ∼ E × B ∼ x̂ × ŷ = ẑ.
So let’s consider a region of area A in the xy-plane. The energy density below the plane was calculated
in (a), and is moving upward (+z) at a speed v = ω/k. The amount of energy that will pass through
the area in a time ∆t is then U = uAv∆t. This is the amount of energy contained in a box with
cross sectional area A and length v∆t. The intensity is the energy per unit area per unit time, so
I = U/(A∆t) = uv = uω/k = 2µω k
E02 e−2κz .

1
3. Griffiths 9.21 Using (9.147) we find
      
 Ẽ 2  1 − β̃ 2 1 − β̃ 1 − β̃ ∗
 0R    µ1 v1
R=  =  = with β̃ = k̃2
 Ẽ0I   1 + β̃  1 + β̃ 1 + β̃ ∗ µ2 ω

Since silver is a good conductor, σ  ω, so as in Problem 9.19(b) above, we have k ≈ κ in the silver, so
 
k̃2 = k2 + iκ2 ≈ κ2 (1 + i) = σωµ2 /2(1 + i). Then β̃ = µ1 v1 σ/2µ2 ω(1 + i) ≡ α(1 + i), where α is collection
of constant, but in particular, is a real number. Then we have



1 − α − iα 1 − α + iα (1 − α)2 + α2
R= =
1 + α + iα 1 + α − iα (1 + α)2 + α2
 
Now we need to evaluate α with the given numbers. α = µ1 v1 σ/2µ2 ω = µ0 c σ/2µ0 ω = 29. Plugging this
into the previous expression we find R = 0.93. So 93% of the light is reflected.
µ0 p20 ω 4
4. Griffiths 11.3 In the text we calculated the total power radiated by a dipole: P  = . If instead there
12πc
2
were a wire connecting the ends of the dipole, the power dissipated as heat would be P = I R. The current we
need to use is the current produced by the oscillating dipole, namely I = −q0 sin(ωt). Thus P = q02 ω 2 sin2 (ωt)R.
Taking the time average, the sine just gives a factor of 12 , so P  = 12 q02 ω 2 R. Setting this equal to the previous

2
µ0 ω 2 d2 2 d
expression for P , recalling that p0 = q0 d, we find R = = πµ0 c , where I used ω = 2πc/λ.
6πc 3 λ
Evaluating this with µ0 = 4π × 10−7 N/A2 and c = 3 × 108 m/s, we get R = 80π 2 (d/λ)2 Ω = 790(d/λ)2 Ω.
For the station 101.1 FM, the frequency is 101.1 MHz, so since c = λf , we find λ ≈ 3 m. Thus d/λ = .005/3,
so R = 0.22 Ω. For AM waves the wavelength is longer, so the radiation resistance is even smaller.

5. Griffiths 11.4 We want to find the radiation due to a single rotating electric dipole, which can be represented
by the superposition of two other electric dipoles, p1 = p0 cos(ωt) x̂ and p2 = p0 sin(ωt) ŷ. In the text we
derived the fields and power due to a single electric dipole oscillating along the z-axis. The superposition
principle guarantees that we can simply add the electric and magnetic fields from two different sources, but it
doesn’t tell us whether we can add the power from the two sources (in general we cannot add powers). So let’s
determine the fields, since we can use superposition on them.
Equation (11.18) in the text gives the E-field for a single electric dipole oscillating along the z-axis. To most
easily use this formula to handle dipoles oscillating along x̂ or ŷ, let’s write sin θ θ̂ in a different way. Since θ
is measured from the z-axis, you can show ẑ = cos θ r̂ − sin θ θ̂. (Remember that θ̂ and r̂ change depending
on where you are in space. So pick a point and draw all three vectors ẑ, r̂, and φ̂ with their tails on that
point.) Since cos θ = z/r, this means: sin θ θ̂ = zr r̂ − ẑ. So now the formula representing a dipole oscillating
along the x-axis will simply have x replacing z. For p2 , we have y replacing z and also the time dependence
also switches from cos to sin. Thus making the proper modifications to (11.18) we have:

µ0 p0 ω 2 x  µ p ω2 y 
0 0
Etot = E1 + E2 = cos(ωt0 ) r̂ − x̂ + sin(ωt0 ) r̂ − ŷ
4πr r 4πr r
µ0 p0 ω 2  x  y 
= cos(ωt0 ) r̂ − x̂ + sin(ωt0 ) r̂ − ŷ
4πr r r
where I’ve used the shorthand t0 = t − r/c. Now, we could do the same thing for B, but it is simpler to notice
that B = 1c ( r̂ × Etot ). We can then more easily calculate the Poynting vector.

1 1 1  2  1 2
S= (E × B) = (E × ( r̂ × E)) = E r̂ − (E · r̂)E = E r̂
µ0 µ0 c µ0 c µ0 c tot
 
Here we used that fact that E · r̂ = 0 because xr r̂ − x̂ · r̂ = xr − xr = 0, and the same holds for the other piece
2
of Etot . So we just need to calculate Etot = (E1 + E2 ) · (E1 + E2 ) = E12 + E22 + 2E1 · E2 . Here’s where the time
2
averaging helps. All the time dependence is in the trig functions. The three terms in Etot are proportional to

2
 cos2 (ωt0 ), sin2 (ωt0 ), and cos(ωt0 )sin(ωt0 ) respectively. The cos2 and sin2 both average to 1
2, while the cross
term averages to 0. So now we have

2   2 1  y 2 
1 µ0 p0 ω 2 1 x
S = r̂ − x̂ + r̂ − ŷ r̂
µ0 c 4πr 2 r 2 r
Working out the term in square brackets we get


1x 2 1  y 2 1 x2 x2 y2 y2 1 x2 + y 2 1
r̂ − x̂ + r̂ − ŷ = 2
− 2 2
+ 1 + 2
− 2 2
+ 1 = 1 − 2
= 1 − sin2 θ
2 r 2 r 2 r r r r 2 r 2
Putting this back, we get the final result

2

µ0 p0 ω 2 1
S = 1− sin2 θ r̂
c 4πr 2
The intensity profile is an ellipsoid centered at the origin which is twice as long in the z-direction as in the x
and y directions.
To find the total power we integrate the intensity over a spherical surface.

2 
  π  
µ0 p0 ω 2 1 1 2 2 µ0 p20 ω 4 1 π 3
P = S · da = 1 − sin θ r sin θ dθdφ = 2π sin θ dθ − sin θ dθ
c 4π r 2 16π 2 c 0 2 0


µ0 p20 ω 4 14 µ0 p20 ω 4
= 2− =
8πc 23 6πc
Note that this is in fact just twice the power emitted by a single oscillating dipole. We could have just added
the power from both dipoles in this case because they were out of phase, which made the cross term ( i.e.
sin(ωt) cos(ωt)) average to zero.

6. Griffiths 11.9 The ring possesses an electric dipole moment which is rotating. There might also be a changing
magnetic dipole moment, but that contribution is much smaller than the contribution from the electric dipole
radiation. (See the discussion at the end of Section 11.1.3.) So we need to first calculate the ring’s electric
dipole moment. Let’s calculate it at a fixed time, t = 0, so it isn’t moving while we integrate.
  
 
p = ρ(r )r dτ = λ(r)r dl = (λ0 sin φ)(b sin φ ŷ + b cos φ x̂) b dφ

 2π  2π 
2 2
= λ0 b ŷ sin φ dφ + x̂ sin φ cos φ dφ = λ0 b2 (π ŷ + 0 x̂) = πb2 λ0 ŷ
0 0

Now we can use the result from the previous problem to find the power radiated by this rotating dipole. We
have p0 = πb2 λ0 . Thus P = µ0 πb4 λ20 ω 4 /(6c).

7. Griffiths 11.14 Treating the hydrogen atom classically, we have an electron orbiting around the proton due
to their electrostatic attraction. Thus we can find the velocity of the electron by setting the Coulomb force
equal to the centripetal force needed to keep the electron in a circular orbit.

1 e2 v2 e2
FCoul = Fcent ⇒ = m e ⇒ v =
4π0 r2 r 4π0 me r

For an initial radius of r = a0 = 5 × 10−11 m, you can plug in numbers and should find v/c = 0.0075. Since
v depends on the square-root of r, when r is 100 times smaller, v/c will only be 10 times bigger, which is still
pretty small. So it is indeed a good approximation to assume that the electron is non-relativistic for most of
its trip.
Because of the centripetal acceleration, a = v 2 /r, the electron will radiate. Using the Larmor formula,

2
2
2
µ0 e2 v 2 µ0 e2 1 e2 e2 1 e2
P = = = .
6πc r 6πc 4π0 mr2 6π0 c3 4π0 mr2

3
 As the electron spirals in, it gains kinetic energy and loses potential energy, and also loses some energy to
radiation. For the total (kinetic plus potential plus radiated) to be conserved, the decrease in the kinetic plus
potential energy must equal the power radiated.

1 1 e2 1 1 e2 1 e2 1 e2
U = UKE + UP E = mv 2 − = − =−
2 4π0 r 2 4π0 r 4π0 r 8π0 r
Then
2
2
dU 1 e2 dr e2 1 e2 dr 4 1 e2 1 A
− =− = P = ⇒ =− 3 ≡− 2
dt 8π0 r2 dt 6π0 c3 4π0 mr2 dt 3c 4π0 m r2 r
Now we can integrate to find the total time.
  a
2
2
2
0
2
T
r3  0 a30 a30 3c3 4π0 m 3 2π0 mc
−r dr = A dt ⇒ − r dr = A dt ⇒ = = AT ⇒ T = = a0 c
a0 0 3 0 3 3 4 e2 e2

Plugging in the values of the constants and double checking that the units work out, you should find T =
1.3 × 10−11 s. Clearly this theory has a problem, because we know the hydrogen atom lasts much longer than
that! This is one reason quantum mechanics was needed.

8. Griffiths 11.21

  p0 = qd. The frequency of oscillation is ω =
(a) We have an oscillating electric dipole with magnitude k/m.
µ p2 ω 4
sin θ 2
The time averaged Poynting vector is S = 32π 0 0
2c r 2 r̂. Let’s choose our z-axis to be pointing down,
towards the floor. The radiation is traveling radially away from our radiating dipole, which means it hits
the floor with some angle at radius R. We can get the power per unit area on the floor by dotting S with
ẑ, the normal to the floor, which gives us a factor of cos θ, where θ is the angle between the vertical line
from the dipole to the floor and the radial line from the dipole to the point at radius R. From this right
triangle we find sin θ = R/r, cos θ = h/r, and r2 = R2 + h2 . So we have:




µ0 p20 ω 4 sin2 θ cos θ µ0 q 2 d2 ω 4 R2 h µ0 q 2 d2 ω 4 R2 h
Ifloor = S · ẑ = 2 2
= 2 5
= 2
32π c r 32π c r 32π c (R2 + h2 )5/2
To find the location of most intense radiation, we take a derivative of I with respect to R and set it equal
to zero.
 
dI d R2 2R 5 2RR2
=0 ⇒ 2 2 5/2
=0 ⇒ 2 2 5/2
− =0 ⇒ R=0
dR dR (R + h ) (R + h ) 2 (R + h2 )7/2
2

5 2
or (R2 + h2 ) = R2 ⇒ R2 = h2
2 3
Since the intensity is zero for R = 0, and falls off to zero as R → ∞, there must indeed be a maximum

for R = 2/3h.
(b) Now we want to find the average energy per unit time (i.e. power) striking an infinite floor. So we need
to integrate I over all R from 0 to ∞.
  ∞
  ∞
µ0 q 2 d2 ω 4 R3 dR
P = I(R) da = I(R)2πR dR = 2π 2
h
0 32π c 0 (R + h2 )5/2
2

This integral can be done using a trig substitution, R2 = h2 tan2 u.

 ∞  
R3 dR 1 π/2
tan3 u 1 π/2
2
= du = sin3 u du =
0 (R + h2 )5/2
2 h 0 sec3 u h 0 3h
2 2 4
Thus the total power that hits the floor is P = µ0 24πc
q d ω
. As you would expect, this is half of the total
power radiated by an electric dipole (compare Eqn. 11.22), because the other half of the radiation hits
the ceiling.

4
(c) If the amplitude of oscillation is x0 (t), then the total energy of the simple harmonic oscillator at any time
is U = 12 kx0 (t)2 . (So x0 (0) = d.) The decrease in energy should be equal to the total power radiated,
which is twice the result found in part b.

dU 1 d µ0 q 2 ω 4 d µ0 q 2 ω 4
− = − k (x0 (t)2 ) = 2P = x0 (t)2 ⇒ (x0 (t)2 ) = − (x0 (t)2 ) = −A(x0 (t)2 )
dt 2 dt 12πc dt 6πkc
⇒ x0 (t)2 = d2 e−At ⇒ x0 (t) = d2 e−At/2

12πkc 12πcm2

Thus when t = 2/A, we’ll have x0 = de−1 = d/e. Thus τ = µ0 q 2 ω 4 = µ0 q 2 k . (Remember, ω = k/m.)

5
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink)
Problem Set 11

1. The electric vector of a fully polarized plane where M is a 2 × 2 Jones matrix representing the
EM wave is given by the expression polarizer. Calculate M . (As usual, the beam di-

 rection is z, φ is an angle in the xy plane, and φ
E = E0 x̂ cos (kz − ωt) + ŷ b cos (kz − ωt + φ) , is positive as one rotates from x̂ toward ŷ.)

where E0 , k, ω, b, and φ are real constants. 3. A wave plate consists of a single crys-
(a) tal in which plane EM waves that are linearly
Defining polarized in the “slow” (“fast”) direction propa-
  gate with phase velocity c/nslow (c/nfast ), where
E ≡ Re Ẽ exp (i(kz − ωt)) , nslow > nfast due to lack of cubic symmetry in the
crystal lattice. Most interesting is the quarter-
show that the first equation is equivalent to wave plate, which has a thickness D such that
 
Ẽ = E0 x̂ + ŷ beiφ . kD(nslow − nfast ) = π/2 .

(b) Usually a quarter-wave plate is deployed with its

Sketch the locus of E/E0 in the x-y plane for slow axis bisecting the x̂ and ŷ axes, where ẑ
the following cases: is the direction of wave propagation; the plate’s
φ = 0, b = 1 fast axis is perpendicular to its slow axis.
(a)
φ = 0, b = 2 Calculate the Jones matrix M that characterizes
φ = π/2, b = −1 this quarter-wave plate (constant multiplicative
phase factors are unimportant).
φ = π/4, b = 1
(b)
(c) Starting with light that is unpolarized, i.e. in
In cases where Ex is nonzero, define the complex which there is no fixed phase relationship be-
constant α according to tween Ẽx and Ẽy , it is possible to obtain fully
    circularly polarized light using only a quarter-
Ẽx 1 1 wave plate and a linear polarizer. Specify in
∝ .
Ẽy 1 + |α|2 α which order these elements should be traversed
by the beam. Using the Jones matrix that you
The right-hand side of this equation (including calculated for the quarter-wave plate, prove that
the normalizing factor) is called the Jones vec- your design will work.
tor J. Write the Jones vectors for the four waves
described in (b). 4. Using a combination of optical elements (lin-
ear polarizer or wave plate), design a system
2. Consider an ideal linear polarizer with its that will pass right-hand circularly polarized
transmission axis at an arbitrary angle φ with light without changing its polarization, but will
respect to the x axis. It acts on a fully po- completely block left-hand circularly polarized
larized plane EM wave that is characterized by light. This system is called a right-hand circu-
initial and final Jones vectors J and J
(only J lar analyzer. Use Jones matrices to prove that
is normalized), such that your design will work. (According to the usual
convention, if you take a snapshot of a right-
J
≡ MJ , hand polarized EM wave, the electric field vec-
tor traces the thread pattern of a right-handed of oppositely vibrating dipoles. Then for z > 0
screw pointed along ẑ.) the resulting diffraction pattern is merely the ra-
diation pattern of the inserted line of oppositely
5. Griffiths 9.31 vibrating dipoles (everything else cancels out).
(This argument is due to Babinet.)
6. Show that the characteristic impedance Consider N equally spaced slits of the type just
Z0 ≡ ∆V /I of the coaxial cable in the previ- described. Put the observer at x = y = 0,
ous problem is z = ∞; direct the incident beam at an angle
 ψ = arctan (ky /kz ) to the z axis. The observer
1 µ b
Z0 = ln sees the radiation from N lines of dipoles – but
2π  a the phase of each line of dipoles differs from that
and that this result is equivalent to of the next line by kb sin ψ, where b is the slit
spacing and k is the wave vector’s magnitude.

L
Calculate the diffraction pattern
Z0 =
C

S(ψ)/S(ψ = 0) ,
where L
and C
are the cable’s inductance and
capacitance per unit length, respectively. where S is the magnitude of the Poynting vector
seen by the observer.
7. Prove that
N
sin N ∆φ/2
exp (iφn ) = exp (iφ̄),
n=1
sin ∆φ/2

where
∆φ ≡ φn+1 − φn ,
and φ̄ is the average of the φn .

8. The result of the previous problem can be

used to calculate the diffraction pattern of an
N -slit system.
Consider a very thin, perfectly conducting screen
at z = 0, upon which a plane EM wave is in-
cident from z < 0, linearly polarized along x̂.
Macroscopically, we know that the wave is fully
reflected; nothing is transmitted. Microscopi-
cally, free electrons in the screen are set into
vibration by the incident electric field. Taken
individually, these dipoles would radiate in the
usual electric dipole pattern. However, when the
effects of all the dipoles are combined, along +ẑ
they radiate a plane wave that exactly cancels
the incident wave; and along −ẑ they radiate a
plane reflected wave.
Now cut a narrow slit in the screen, along
z = 0, y = constant. This is equivalent to re-
moving a line of dipoles – or to inserting a line
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink )

Solution Set 11 (by Daniel Larson)

1. Polarization

(a) We want to convert the real electric field E into a complex electric field.



  
E = E0 Re ei(kz−ωt) x̂ + E0 bRe ei(kz−ωt+φ) ŷ = Re E0 x̂ + ŷbeiφ ei(kz−ωt)
 
iφ 1
Ẽ = E0 ( x̂ + be ŷ) =
beiφ
(b) The real E-field vector lives in the xy-plane. We want to sketch the path traced by the tip of the E-field
for fixed z as we let the time increase. The easiest way to go about determining the path is to set z = 0,
π π
and then use the values for b and φ in the real vector E and plot some sample points for t = 0, 2ω , ω , . . ..

y y

x x

b=1 φ=0 ( 11) b=2 φ=0

( 12)
y y

x x

_
b=−1 φ=π/2 ( ) 1 b=1 φ=π/4 ./2
( )
−i 1+i

These plots represent light with linear polarization (top two), right-hand circular polarization (lower left)
and left-hand elliptical polarization (lower right).
 
1 1
(c) The Jones vectors are given by √ ; the un-normalized vectors are shown next to the plots
1 + b2 beiφ
above. When properly normalized they are, respectively:
       
1 1 1 1 1 1 1 1
√ √ √ √ 1
2 1 5 2 2 −i 2 √2 (1 + i)

1
2. Polarizer A Jones vector J represents the polarization of a beam of light. When that beam passes through a
polarizer, the polarization is changed. We would like to be able to represent the effect of the polarizer by a 2 × 2
matrix that acts on J representing the initial beam and gives back J representing the final beam: M J = J .
We want to find the matrix M for a polarizer aligned with its transmission axis at an arbitrary angle φ with
respect to the x-axis.
Let’s start by solving a simpler system, where the transmission axis is along the xA -axis. We’re going to change
basis later, so I want to label this x − y coordinate system with the subscript A. All the vector coordinates
and matrices given with respect to this basis will also have a subscript A. Incoming light with x-polarization

is represented by the Jones vector JA = 10 A , and it will pass through the polarizer unchanged. (It’s traveling

along the “transmission” axis, after all.) On the other hand, light with y-polarization, JA = 01 A , will be
completely blocked. These conditions are enough to specify the matrix MA in the A-basis.
         
1 1 0 0 1 0
MA = MA = ⇒ MA =
0 A 0 A 1 A 0 A 0 0

Now we want to change basis to a new one (with coordinates xB and yB ) such that the transmission axis of
the polarizer (the xA -axis) makes an angle of φ with the new xB -axis. Let’s let the matrix that accomplished
this change of basis be called R, so that a Jones vector expressed in the B-basis, JB , will be expressed as the
vector JA = RJB in the A-basis. We can also go back the other way: JB = R−1 JA . What we really want is
the matrix MB , where MB JB = JB . In terms of MA and R we find:

MA JA = JA ⇒ MA RJB = RJB ⇒ R−1 MA RJB = JB ⇒ MB = R−1 MA R

So now all we need to do is determine R. Consider a unit vector on the xB -axis. In the B-coordinates it is
   1  cos φ 
given by 10 B while in the A-coordinates it is given by −cos φ
sinφ A . Thus R 0 B= − sin φ B . You can do the
 cos φ sin φ
same analysis starting with 01 B . The matrix R is then R =
− sin φ cos φ
      
−1 cos φ − sin φ 1 0 cos φ sin φ cos2 φ sin φ cos φ
MB = R MA R = =
sin φ cos φ 0 0 − sin φ cos φ sin φ cos φ sin2 φ
A B

3. Wave Plate

(a) We can use the same method to find the matrix for the quarter-wave plate. Namely, first solve the problem
in an easier situation, where the slow axis is along the xA -axis, and then change to the B-coordinates
where the slow axis is at 45◦ with respect to the xB axis. We can use the same R matrix we found above,
with φ = 45◦ .

So we first need to find MA . If we have light polarized along the slow axis, JA = 10 A , it will pass through
the quarter-wave plate and pick up some arbitrary phase that we might as well take to be zero. Then

when light polarized along the fast axis, JA = 01 A , passes though, it picks up a quarter-wavelength less

phase, so it is π/2 behind the wave along the slow axis. So if the slow-axis wave emerges with JA = 10 A ,
 0  0
the fast-axis wave will emerge with Jones vector JA = e−iπ/2 = −i . This means the matrix for
  A
1/4 1 0
the quarter-wave plate is given by MA = in the A-coordinates. Using R and R−1 with
0 −i
A
φ = π/4, we find:
 √ √    √ √   
1/4 −1 1/4 1/ 2 −1/ 2 1 0 1/ 2 1/ 2 1 1−i 1+i
MB = R M A R = √ √ √ √ =
1/ 2 1/ 2 0 −i −1/ 2 1/ 2 2 1+i 1−i
A B

2
 This might not be exactly what you expected, but remember we can multiply the whole matrix by an
overall phase without changing anything. Choosing the phase eiπ/4 = √12 (1 + i) we find
   
1 1−i 1+i 1 1 i
MB = eiπ/4 =√
2 1+i 1−i 2 i 1
B B

(b) Now we can use these matrices to design optical systems. We want to take unpolarized light and make it
1/4   
circularly polarized. First notice that MB 10 = √12 1i , namely the quarter wave plate turns x-polarized
light into LH circularly polarized light. Thus by first using a linear polarizer aligned along the x-axis,
we can turn the unpolarized beam into an x-polarized beam, which can then pass through the quarter
wave plate and become LH circularly polarized. In terms of matrices, our system of x-polarizer then
quarter-wave plate is represented by the matrix
    
1 1 i 1 0 1 1 0
Mtot = M 1/4 M x = √ =√
2 i 1 0 0 2 i 0

Note that the order of matrices is “opposite” the order in which the optical elements are traversed (as
reading from left to right). This is because it is the matrix on the right that is the first to act on any
 
Jones vector. To test if this matrix does what we want, multiply it by any random vector. Mtot ab =
  
√1 a = √a 1 , which is indeed LH circularly polarized.
2 ai 2 i

4. Circular Analyzer Now we want to make a RH circular analyzer that passes RH circularly polarized light
unchanged and completely blocks LH circularly polarized light. The idea is to use a quarter-plate to make
circularly polarized light into linearly polarized light, then use a linear polarizer to allow only the component
you want through, and then use the quarter-wave plate again to make the linearly polarized light circular again.
The important observation is that a quarter-wave plate turns x-polarized light into LH circularly polarized,
LH circular into y-polarized, y-polarized into RH circular, and RH-circular into x-polarized. (This is easier to
visualize if you draw a little flow chart.) So we will send a light beam through the quarter-wave plate. The
RH components will turn into x-polarized light and the LH components will become y-polarized. Then we use
a linear polarizer with the x-axis as its transmission axis, so only x-polarized light gets through. Then we can
use 3 quarter-wave plates, or equivalently one quarter-wave plate rotated by pi, to convert the x-polarized light
back to RH circular. In matrices this becomes:
      
1 −1 i 1 0 1 1 i 1 1 i
(M 1/4 )3 M x M 1/4 = √ √ =−
2 i −1 0 0 2 i 1 2 −i 1
 
Now to test it, multiply by a the vector J = ab , which gives (dropping the irrelevant overall minus sign)

1 a+ib
 1
 1
 1
2 b−ia = 2 (a + ib) −i . Now, RH circularly polarized light has a = 1, b = −i, so the output is −i which is
RH circular, just as we want. On the other hand, LH circular light has a = 1, b = i which yields 0, so it is
completely blocked. This design works as desired!

5. Griffiths 9.31

(a) The first part of the problem consists of plugging equations (9.197) into (9.177) and (9.175). To take the
derivatives in Maxwell’s equations we need the formulas for divergence and curl in cylindrical coordinates,
as found in the front cover.
   
A cos(kz − ωt) 1 ∂ A cos(kz − ωt) 1 ∂
∇· E = ∇· ŝ = s =0 ∇· B = (Bφ ) = 0
s s ∂s s s ∂φ

∂Es 1 ∂Es 1 ∂B Aω
∇× E = φ̂ − ẑ = − Ak sin(kz − ωt) φ̂ while − =− sin(kz − ωt) φ̂
∂z s ∂φ s ∂t cs

3
 The above two terms are equal because ω = ck. Similarly,
∂Bφ 1 ∂ Ak 1 ∂E Aω
∇× B = − ŝ + (sBφ ) ẑ = sin(kz − ωt) ŝ while = 2 sin(kz − ωt) ŝ
∂z s ∂s cs c2 ∂t c s
The boundary conditions are easily satisfied: E  = Ez = 0 and B ⊥ = Bs = 0.
(b) To find the charge density, we can use Gauss’s Law with a cylindrical surface with radius s and length

dz enclosing the inner cylinder. E · da = As cos(kz − ωt)(2πs)dz = Qenc /0 = λdz/0 ⇒ λ =
2π0 A cos(kz − ωt).
To determine I we use an Amperian loop of radius s. We need to take into account both the “regular”
current and also the displacement current. However, since the displacement current is 0 ∂E ∂t , and E is
A
only radial, none of it will pass through a loop enclosing the inner cylinder. Thus B · dl = cs cos(kz −
2πA λ
ωt)(2πs) = µ0 Ienc ⇒ I = µ0 c cos(kz − ωt). Note that I = 0 µ0 c .

6. Characteristic Impedance The characteristic impedance is defined as Z0 ≡ ∆V /I. In Problem 6 of Home-

work #2 we calculated the potential difference and capacitance per unit length for a system of coaxial cylinders.
λ
We found ∆V = 2π 0
ln(b/a), and C  = 2π0 / ln(b/a). So
 

∆V λ ln( ab ) 0 µ0 c 1 µ0 b 1 µ0 b
Z0 = = = √ ln = ln
I 2π0 λ 2π µ0 0 a 2π 0 a

µ0
b
The inductance per unit length for this arrangement was calculated in Example 7.13 in the text. L = 2π ln a .
Thus
   b
 
L µ0 b ln a 1 µ0 b
= ln = ln = Z0
C 2π a 2π0 2π 0 a

7. Identity The first important thing to notice is that since φn − φn−1 = ∆φ for all n, we can write each φn in
terms of φ1 and ∆φ: φn = φ1 + (n − 1)∆φ. Expanding the sum and using this relation, we get:
N

eiφn = eiφ1 + eiφ2 + · · · + eiφN = eiφ1 + ei(φ1 +∆φ) + ei(φ1 +2∆φ) + · · · + ei(φ1 +(N −1)∆φ)
n=1

= eiφ1 1 + ei∆φ + e2i∆φ + · · · + ei(N −1)∆φ

The term in parentheses looks like a geometric series. Notice that (1 + a + a2 + · · · + aN −1 )(1 − a) = 1 − aN ⇒
(1 + a + a2 + · · · + aN −1 ) = (1 − aN )/(1 − a). Using this identity with a = ei∆φ , we get
N
 1 − eiN ∆φ iφ1 1 − e
iN ∆φ −iN ∆φ/2 iN ∆φ/2
e e iφ1 iN ∆φ/2 −i∆φ/2 e
−iN ∆φ/2
− eiN ∆φ/2
eiφn = eiφ1 = e = e e e
n=1
1 − ei∆φ 1 − ei∆φ e−i∆φ/2 ei∆φ/2 e−i∆φ/2 − ei∆φ/2
sin(N ∆φ/2)
= ei(φ1 +(N −1)∆φ/2)
sin(∆φ/2)

We’re getting close. Let’s step aside and calculate φ̄.

N N
 
1  1  1  1 N (N − 1) ∆φ
φ̄ = φn = (φ1 +(n−1)∆φ) = N φ1 + ∆φ N (n − 1) = φ1 + ∆φ = φ1 +(N −1)
N n=1 N n=1 N n=1
N 2 2

How nice! This is exactly what is in the exponent of the phase factor above. Thus

N sin N ∆φ2
eiφn =  eiφ̄
n=1 sin ∆φ
2

4
8. Diffraction We have a screen with N identical long thin slits cut in it. First let’s deal with the situation when
ψ = 0; i.e. the incoming wave is normal to the screen. Let E0 be the electric field at the location of the observer
due to one of the slits. Since the slits are identical, and the light from each of them is in phase with the others,
each one contributes E0 to the total E-field seen by the observer. Thus Etot = E0 + E0 + · · · + E0 = N E0 .
 2
The magnitude of the Poynting vector measured by the observer is S =  µ10 c Etot  = µ10 c N 2 |E0 |2 .
Now consider the case when the incident wave hitting the screen with the slits is at an angle ψ. This means
that the electric field from each of the slits will be out of phase from the one below it by ∆φ = kb sin ψ. Let us
define the overall phase so that the E-field from the first slit is E0 eiφ1 . Then

N sin N ∆φ2
Etot = E1 + E2 + · · · = E0 eiφ1 + E0 eiφ2 + · · · + E0 eiφN = E0 eiφn = E0 eiφ̄ 
n=1 sin ∆φ
2

Now we can compute S. 

2 ∆φ
1 1 sin N 2
S(ψ) = |Etot |2 = |E0 |2 
µ0 c µ0 c 2 ∆φ
sin 2

(The complex exponential has a modulus of one: |eiφ̄ |2 = 1.) Now when we make the ratio S(ψ)/S(0), the
constants and |E0 |2 pieces cancel, leaving just an N 2 from the denominator. Using ∆φ = kb sin ψ we find:
 
S(ψ) sin2 12 N kb sin ψ
= 2 2 1 
S(ψ = 0) N sin 2 kb sin ψ

If we evaluate this result for ψ = 0, using the small angle approximation on the outer sine functions, we get
S(0)/S(0) = 1, as it must.

5
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink)
MIDTERM EXAMINATION 1

Directions: Do all 3 problems, which have unequal weight. This is a closed-book closed-note exam
except for one 8 12 × 11 inch sheet containing any information you wish on both sides. A photocopy
of the four inside covers of Griffiths is included with the exam. Calculators are not needed, but you
may use one if you wish. Laptops and palmtops should be turned off. Use a bluebook. Do not use
scratch paper – otherwise you risk losing part credit. Cross out rather than erase any work that
you wish the grader to ignore. Justify what you do. Express your answer in terms of the quantities
specified in the problem. Box or circle your answer.

Problem 1. (45 points)

A surface charge of uniform density σ0 Coul/m2
is glued onto a spherical shell of radius R that is
centered at the origin.
(a) (10 points)
Relative to ∞, find the potential V0 at the
origin.
(b) (5 points)
How much work W was done to move the charge
from ∞ to the shell?
(c) (10 points)
The shell is now split along its “equator” into two
hemispheres, and the south hemisphere is thrown
away. Find the new potential V1/2 at the origin.
(d) (20 points)
For the conditions of part (c), calculate the
potential VN at the “north pole” (0, 0, R).

Problem 2. (25 points)

A point charge q is held at a distance z above
an infinite conducting plane that is grounded
(V = 0). Calculate the surface charge density
σs on the plane at a distance s  z from the
charge. Accuracy to lowest nonvanishing order
in z/s is sufficient.

Problem 3. (30 points)

A thin phonograph record is composed of a ma-
terial that has a uniform volume charge density;
the total charge is Q. The record has radius R
and rotates on a turntable at angular velocity ω
.
Calculate the magnetic field at the center of the
record.
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink)
SOLUTION TO MIDTERM EXAMINATION 1

Directions: Do all 3 problems, which have unequal weight. This is a closed-book closed-note exam
except for one 8 12 × 11 inch sheet containing any information you wish on both sides. A photocopy
of the four inside covers of Griffiths is included with the exam. Calculators are not needed, but you
may use one if you wish. Laptops and palmtops should be turned off. Use a bluebook. Do not use
scratch paper – otherwise you risk losing part credit. Cross out rather than erase any work that
you wish the grader to ignore. Justify what you do. Express your answer in terms of the quantities
specified in the problem. Box or circle your answer.

Problem 1. (45 points) hemispheres, and the south hemisphere is thrown

A surface charge of uniform density σ0 Coul/m2 away. Find the new potential V1/2 at the origin.
is glued onto a spherical shell of radius R that is Solution:
centered at the origin. We could have obtained the answer to (a) by
(a) (10 points) doing the integral
Relative to ∞, find the potential V0 at the ori-


gin. ρ(r
)
4π0 V0 = dτ
.
Solution: r

This part of the problem is spherically symmet-
ric. Outside the shell, the potential is that of a Now, with half of the shell removed, the integral
point charge. Inside the shell, there is no charge, is half as big. Therefore
so the potential there is the same as at the shell’s
surface. Therefore V0 σ0 R
V1/2 = = .
q 2 20
4π0 V0 =
R
4πR2 σ0 (d) (20 points)
=
R For the conditions of part (c), calculate the po-
σ0 R tential VN at the “north pole” (0, 0, R).
V0 = .
0 Solution:
Now we need actually to do an integral. Con-
(b) (5 points) sider a ring dθ
of charge, where θ
is the angle
How much work W was done to move the charge measured from the north pole. This ring has
from ∞ to the shell? area da
= 2πR2 sin θ
dθ
and is located a dis-


Solution: tance r
= 2R sin θ2 from the north pole. The

contribution from this ring to the potential at
W = 2 dτ
ρ(r
)V (r
)
1 the north pole is

= 12 qV (R) σ0 da

4π0 dVN =
σ0 R r

= 12 4πR2 σ0
0 σ0 2πR2 sin θ


= dθ
2πR3 σ02 2R sin θ2
W = .
0 πRσ0 sin θ


=
dθ .
(c) (10 points) sin θ2
The shell is now split along its “equator” into two
Substituting Apart from factors of order unity, the answer

−qz/s3 could be guessed. Since a dipole is in-
θ

sin θ
= sin ( θ2 + 2) volved, the result must be proportional to its
= 2 sin θ

θ
moment and thus to z. Given that, −qz/s3 is
2 cos 2 ,
the only acceptable combination of the available
we have variables that has the dimensions of a surface


charge density. This argument is worth some
πRσ0 2 sin θ2 cos θ2
4π0 dVN =
dθ
part credit.
sin θ2


This problem could also be approached by con-
= 2πRσ0 cos θ2 2d θ2 . sidering separately the electric fields from the
π physical and image charges, expanding them in
Integrating over 0 < θ
< 2, powers of z/s, and retaining the leading terms

π/4 that do not cancel. If you attempted to do this
4π0 VN = 4πRσ0 cos θ2 d θ2 and fouled it up, you shouldn’t expect excessive
0
part credit, as such an approach doesn’t require
= 4πRσ0 sin π4 excessive physical insight.
σ0 R
VN =√ .
20
As a check, if we had integrated θ
all the way
to π, including both hemispheres, we would have
recovered the answer to (a).

Problem 2. (25 points)

A point charge q is held at a distance z above
an infinite conducting plane that is grounded
(V = 0). Calculate the surface charge density
σs on the plane at a distance s  z from the
charge. Accuracy to lowest nonvanishing order
in z/s is sufficient.
Solution:
For z > 0, the effect of the charge that is in-
duced on the conducting plane is the same as
that of an image charge −q a distance z below
the plane. Together the physical charge and the
image charge form a physical dipole with mo-
ment p = ẑq2z. At a cylindrical radius s  z,
the field of the physical dipole is approximately
the same as that of an ideal dipole:
4π0 r3
E = 3r̂(r̂ · p̂) − p̂
p
= 3ŝ(ŝ · ẑ = 0) − ẑ
2qz
4π0 E = −ẑ 3 .
s
The surface charge density on the conductor is
just 0 Ez , so
qz
σs = − .
2πs3
Problem 3. (30 points)
A thin phonograph record is composed of a ma-
terial that has a uniform volume charge density;
the total charge is Q. The record has radius R
and rotates on a turntable at angular velocity ω
.
Calculate the magnetic field at the center of the
record.
Solution:
Again we need to do an integral. Define ẑ ≡ ω̂
and s to be the (cylindrical) radius. Consider
an element ds of the record, located a distance s
from its center. The charge dQ on this element is

2πs ds
dQ = Q
πR2
2Qs
= 2 ds .
R

This charge rotates once every 2π/ω seconds, so

the element carries a current
ω 2Qs
dI = ds
2π R2
ωQs
= ds .
πR2
When one applies the Biot-Savart law, one finds
that a circular loop of current I0 and radius s0
has a central field equal to µ0 I0 /2s0 . Therefore
the contribution of the record element ds to the
central magnetic field is
µ0
dB = ẑ dI
2s
µ0 ωQs
= ẑ ds
2s πR2
µ0 ωQ
= ẑ ds
2πR2


µ0 ωQ R
B = ẑ ds
2πR2 0
µ0 ωQ
B = ω̂ .
2πR
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink)
MIDTERM EXAMINATION 2

Directions: Do both problems, which have equal weight. This is a closed-book closed-note exam
except for two 8 12 × 11 inch sheets containing any information you wish on both sides. A photocopy
of the four inside covers of Griffiths is included with the exam. Calculators are not needed, but you
may use one if you wish. Laptops and palmtops should be turned off. Use a bluebook. Do not use
scratch paper – otherwise you risk losing part credit. Cross out rather than erase any work that
you wish the grader to ignore. Justify what you do. Express your answer in terms of the quantities
specified in the problem. Box or circle your answer.

Problem 1. (50 points) (c) (15 points)

A cylindrically symmetric region is bounded by If you were asked to calculate the current density
−∞ < z < ∞ and s < s0 (s is the cylindrical J for the conditions of part (b), where A decays
radius in Griffiths’ notation). Within this re- with time, would you expect J to have the same
gion, the magnetic field may be obtained from dependence on s within our cylindrical region
the vector potential that you obtained in part (a)? Why or why not?

A(s) = ẑµ0 Cs2 , Problem 2. (50 points)

A nickel (five-cent coin) of radius a and thickness
where C is uniform, i.e. independent of r. (You d a carries a uniform permanent magnetiza-
don’t need to choose a particular gauge in order tion
to work this problem, but, if it is helpful, you M = ẑM0 ,
may work in Lorentz gauge ∇ · A + 0 µ0 ∂V /∂t =
0.) where M0 is a positive constant and ẑ is the
(a) (15 points) nickel’s axis of cylindrical symmetry.
For this part, take C to be a (positive) constant, (a) (30 points)
i.e. independent of time t as well as r. Calculate Calculate the magnetic field B(0, 0, 0) at the cen-
the current density J, flowing within this region, ter of the nickel. The direction of B is important;
that produces A. The direction and sign of your express B to lowest nonvanishing order in d/a.
answer are important. (In this application, note (b) (20 points)
that
In the plane z = 0, draw counterclockwise a
J(r
) large circular loop s = b
a that is centered on
4π
µ0 A(r) 
= dτ
,
|r − r
| the nickel. What magnetic flux Φ flows through
this loop? The sign of Φ is important; express Φ
because the current-carrying region is infinite in
to lowest nonvanishing order in d/b.
extent.)
(b) (20 points)
For this part, take C to be a decaying function
of time, i.e.

C(t) = C0 exp (−t/τ ) ,

where C0 and τ are positive constants. Consider

a rectangular loop drawn at constant azimuth φ,
bounded by 0 < z < z0 and 0 < s < s0 . Cal-
culate the EMF E around this loop (the sign of
your answer won’t be graded).
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink)
SOLUTION TO MIDTERM EXAMINATION 2

Directions: Do both problems, which have equal weight. This is a closed-book closed-note exam
except for two 8 12 × 11 inch sheets containing any information you wish on both sides. A photocopy
of the four inside covers of Griffiths is included with the exam. Calculators are not needed, but you
may use one if you wish. Laptops and palmtops should be turned off. Use a bluebook. Do not use
scratch paper – otherwise you risk losing part credit. Cross out rather than erase any work that
you wish the grader to ignore. Justify what you do. Express your answer in terms of the quantities
specified in the problem. Box or circle your answer.

Problem 1. (50 points) Notice that A and J point in opposite direc-

A cylindrically symmetric region is bounded by tions!
−∞ < z < ∞ and s < s0 (s is the cylindrical The above is the most direct path to the result.
radius in Griffiths’ notation). Within this re- Alternatively, one may first evaluate B:
gion, the magnetic field may be obtained from
the vector potential B=∇×A
A(s) = ẑµ0 Cs2 , = −φ̂
∂Az
where C is uniform, i.e. independent of r. (You ∂s
don’t need to choose a particular gauge in order = −φ̂ 2µ0 Cs ,
to work this problem, but, if it is helpful, you
may work in Lorentz gauge ∇ · A + 0 µ0 ∂V /∂t = where the term in the middle equation includes
0.) the only nonvanishing derivative in the curl.
(a) (15 points) Then
µ0 J = ∇ × B
For this part, take C to be a (positive) constant,
i.e. independent of time t as well as r. Calculate ẑ ∂
= sBφ
the current density J, flowing within this region, s ∂s
that produces A. The direction and sign of your =−
ẑ ∂
s2µ0 Cs
answer are important. (In this application, note s ∂s
that
J = −ẑ 4C .
J(r
)
4π
µ0 A(r) 
= dτ
, (b) (20 points)
|r − r
|
because the current-carrying region is infinite in For this part, take C to be a decaying function
extent.) of time, i.e.
Solution:
Combining Ampère’s law with Griffiths’ vector C(t) = C0 exp (−t/τ ) ,
identity (11),
where C0 and τ are positive constants. Consider
µ0 J = ∇ × B a rectangular loop drawn at constant azimuth φ,
= ∇ × (∇ × A) bounded by 0 < z < z0 and 0 < s < s0 . Cal-
= ∇(∇ · A) − ∇2 A culate the EMF E around this loop (the sign of
your answer won’t be graded).
1 ∂ ∂
=0− s ẑµ0 Cs2 Solution:
s ∂s ∂s The electric field is easily calculated from
1 ∂ 2
J = −ẑ C 2s
s ∂s ∂A
= −ẑ 4C . E = −∇V − .
∂t
The potential term integrates to zero around the day’s law yields the EMF:
loop and thus plays no role. Because A is in the
ẑ direction and vanishes on the z axis, the only
dΦ
contribution to the integral comes from the outer E =−
dt
segment where s = s0 and dl = ẑdz. Proceeding
µ0 C0 s20 z0 exp (−t/τ )
counterclockwise around the loop, =− .
τ

E= E · dl
(c) (15 points)

∂A If you were asked to calculate the current density
=− · dl
∂t J for the conditions of part (b), where A decays

0 with time, would you expect J to have the same
∂
=− ẑµ0 C0 s20 exp (−t/τ ) · ẑdz dependence on s within our cylindrical region
z0 ∂t

0 that you obtained in part (a)? Why or why not?
µ0 C0 s20 exp (−t/τ ) Solution:
= dz
z0 τ Now that conditions are not static, Maxwell’s
µ0 C0 s20 z0 exp (−t/τ ) corrected version of Ampère’s Law is needed:
=− .
τ
∂E
The above is the most direct path to the result. µ0 J = ∇ × B − µ0 0 .
∂t
Alternatively, one may first calculate the mag-
netic flux Φ through the loop, then obtain E from
its time derivative. This flux is most easily evalu- Though ∇ × B has no s-dependence within our
ated by performing the line integral of A around cylindrical region, the contribution of dE/dt to
the loop. Again proceeding counterclockwise, J is proportional to s2 , as is A itself. Therefore
the Maxwell-corrected J will not have the same

s-dependence as in part (a).
Φ= A · dl

0
Problem 2. (50 points)
= µ0 C0 s20 exp (−t/τ )dz A nickel (five-cent coin) of radius a and thickness
z0
d a carries a uniform permanent magnetiza-
= −µ0 s20 z0 C0 exp (−t/τ ) .
tion
This same flux may also be obtained by integrat- M = ẑM0 ,
ing B from part (a). Proceeding counterclock-
wise around the loop, da is in the φ̂ direction, where M0 is a positive constant and ẑ is the
opposite to the direction of B. Therefore the flux nickel’s axis of cylindrical symmetry.
is negative. Performing the integration, (a) (30 points)
Calculate the magnetic field B(0, 0, 0) at the cen-
B = −φ̂ 2µ0 Cs ter of the nickel. The direction of B is important;


express B to lowest nonvanishing order in d/a.
Φ = B · da Solution:

s0
z0 The volume magnetization M yields a bound
=− ds dz 2µ0 Cs surface current Kb = M × n̂. Therefore Kb van-
0 0
ishes on the nickel’s flat surfaces, and is equal to
= −µ0 s20 z0 C φ̂M0 on its curved surface. A surface current on
= −µ0 s20 z0 C0 exp (−t/τ ) . this thin curved strip d a is equivalent to a
line current Ib = Kb d. Therefore B at the cen-
With the same flux calculated either way, Fara- ter is the same as the field from a circular loop.
Applying the Biot-Savart law, Note that, as b → ∞, all the flux through the
nickel is returned within the large circle, so
4π dl
× (r − r
)
dB(r = 0) = Φ → 0.
µ0 I |r − r
|3 The above is the most direct path to the result.
φ̂sdφ × (−ŝ) An alternative approach starts from the equation
=
s2 
dφ B · da = 0 .
= ẑ
a
µ0 Kb d
B(0) = ẑ Choose a closed surface consisting of the plane
2a
z = 0 plus the hemispherical cap r = ∞. The cap
d
= ẑ µ0 M0 . makes no contribution to the integral because B
2a from a dipole diminishes as r−3 . The plane can
(b) (20 points) be divided into s < b and s > b. Since the surface
In the plane z = 0, draw counterclockwise a integral over the plane vanishes, the inner and
large circular loop s = b
a that is centered on outer portions give equal and oppposite contribu-
the nickel. What magnetic flux Φ flows through tions. We evaluate the outer portion because the
this loop? The sign of Φ is important; express Φ dipole approximation works well in that region.
to lowest nonvanishing order in d/b.
b
2π
Solution: Φ= ds s dφ Bz
Far from the nickel, the field is that of a magnetic 0 0

∞
2π
dipole with moment
=− ds s dφ Bz .
m = ẑ M0 πa2 d . b 0

But the perfect-dipole approximation breaks In the plane z = 0, with m̂ = ẑ, the dipole’s
down when we get close to the nickel, so it’s magnetic field is
tough to calculate Φ by integrating B over the
4πr3
loop’s inner area. B = 3(m̂ · r̂)r̂ − m̂
The most straightforward approach uses the fact µ0 m
that the flux Φ through a loop is the integral of 4πs3
B = 3(ẑ · ŝ)ŝ − ẑ
A around the loop; the dipole approximation for µ0 m
A will work well at the boundary of the loop, = −ẑ
where b
a. First calculate A: µ0 m
B = −ẑ
4π m × r̂ 4πs3
A= µ0 πa2 M0 d
µ0 r2 = −ẑ
M0 πa2 d 4πs3
= ẑ × (ẑ cos θ + ŝ sin θ) µ0 a2 M0 d
r2 = −ẑ .
M0 πa2 d 4s3
= ẑ × ŝ
b2 Performing the integral over the outer region,
M0 πa2 d
= φ̂
∞

 µ a2 M d 
2π
b2 0 0
µ0 M0 a2 d Φ=− ds s dφ − 3
A = φ̂ . b 0 4s
4b2 2
∞
µ0 a M0 d ds
= 2π
Since A is in the azimuthal direction, its line 4 b s2
integral around the large circle is just 2πbA, so d
= µ0 πa2 M0 .
d 2b
Φ = µ0 πa2 M0 .
2b
 University of California, Berkeley
Physics 110A Fall 2001 Section 1 (Strovink)
FINAL EXAMINATION

Directions: Do all six problems, which have unequal weight. This is a closed-book closed-note exam
except for three 8 12 × 11 inch sheets containing any information you wish on both sides. A photocopy
of the four inside covers of Griffiths is included with the exam. Calculators are not needed, but you
may use one if you wish. Laptops and palmtops should be turned off. Use a bluebook. Do not use
scratch paper – otherwise you risk losing part credit. Cross out rather than erase any work that
you wish the grader to ignore. Justify what you do. Express your answer in terms of the quantities
specified in the problem. Box or circle your answer.

Problem 1. (30 points) (d) (5 points)

A conducting sphere of radius a, centered at the B a distance s outside a long thin wire carrying
origin, carries a constant total charge q0 . Out- current I along ẑ.
side it, between radii r = a and r = b, lies a (e) (5 points)
spherical shell composed of an insulating dielec- B at the center (0, 0, 0) of a circular wire loop of
tric with “frozen-in” polarization radius b lying in the xy plane, carrying counter-
clockwise current I.
q0 (f) (5 points)
P = r̂ ,
4πr2 A inside a long circular cylinder with its axis
along ẑ, containing a magnetic field B = B0 ẑ
where q0 is the same constant. inside, and B = 0 outside.

Calculate the electric field E(r) over the entire

region 0 < r < ∞. (Note that the dielectric
constant  is not defined or supplied here, and
should not appear in your answer.) Problem 3. (30 points)
A long thick cylindrical wire of radius b carries a
Problem 2. (30 points). Static fields. steady current I0 along its axis ẑ, uniformly dis-
Please write down the following simple static tributed over the wire’s cross section. At t = 0
electromagnetic fields in vacuum. Credit will the wire is cut with a thin saw to produce a thin
be based entirely upon your answer; to receive gap in the region − d2 < z < d2 , with d  b. Ne-
any credit, your answer must be exactly correct, glect fringing effects near s = b.
including the field’s direction as well as its mag- (a) (15 points)
nitude. For a period of time after t = 0, the power sup-
(a) (5 points) ply that is connected to the distant ends of the
E inside a parallel plate capacitor that holds wire forces the same current I0 to continue to
charges ±q on plates of area a located at y = ± d2 . flow in the wire. Calculate the magnitude and
(Note that the positive plate is on top.) direction of the magnetic field in the gap.
(b) (5 points) (b) (15 points)
E a distance r from the origin, where r is out- At some later time, a resistor is substituted for
side a spherically symmetric distribution of total the power supply, and the charge that accumu-
charge Q that is centered at the origin. lated on the faces z = ± d2 is allowed to drain
(c) (5 points) away. While this charge is draining, would you
E at (x = 0, 0, 0) produced by an ideal electric expect the electric field in the gap to continue to
dipole of moment p = p0 ẑ that is centered at be exactly uniform (independent of s)? Why or
the origin. why not?
Problem 4. (40 points) Problem 6. (40 points)
A nonrelativistic electron of mass m and charge e A waveguide consists of an evacuated rectangu-
moves in vacuum in the xy plane under the influ- lar pipe that runs parallel to the ẑ axis. The
ence of a constant uniform magnetic field B that pipe has three perfectly conducting metal sides,
is directed along the z axis. Because no other ex- at x = 0, x = a, and y = 0. These three sides
ternally applied fields or mechanical forces exist, are connected together in a “U” shape. The
the electron travels very nearly in a periodic or- fourth (top) side, at y = b, is made of the same
bit. After one revolution, it is observed that the material but is insulated from the “U”.
electron’s kinetic energy has diminished slightly,
Operating in the TEM mode, the waveguide car-
by a factor 1 − η, where η  1.
ries an electromagnetic pulse that travels in the
In terms of B and fundamental constants, calcu- +ẑ direction. At z = 0 and t = 0, a snapshot is
late η. taken of the (nonzero) magnetic field B(x, y, z =
0, t = 0). Calculate B(x, y, z = 0, t = 0) within
a multiplicative constant.
Problem 5. (30 points)
A plane wave of wavelength λ is normally in-
cident on a system of thin slits at constant y
in the aperture plane z = 0. An observer at
z = ∞ observes the Fraunhofer-diffracted beam
at a small angle θ ≡ arctan (dy/dz). In each
part of this problem, you are asked to calculate
the diffraction-pattern ratio

I(θ)
R(θ) ≡ ,
I(0)

where the intensity I is proportional to the

square of the wave amplitude, i.e. to the time-
averaged Poynting vector.
(a) (10 points)
Write down R(θ) for two thin slits at y = ±a/2.
(b) (10 points)
Write down R(θ) for four thin slits, two at

y = +(a ± b)/2 ,

and two at
y = −(a ± b)/2 ,

where a > b > 0.

(c) (10 points)
Take the incident beam to be circularly polar-
ized. Repeat part (b) under the same conditions,
except that an x̂ polarizer is placed behind the
top pair of slits, and an otherwise identical ŷ
polarizer is placed behind the bottom pair.