Massachusetts Institute of Technology

Department of Physics

Physics 8.07 Fall 2005

Problem Set 9 Solutions

Problem 1: Griffiths Problem 8.12 (p. 362)

Place the electric charge at ~x = 0 and the magnetic charge at ~x = d~ez . Then

~ x) = 1 qe ~ x ) = µ0 qm
E(~ ~
x , B(~ (~x − d~ez ) .
4π0 r3 4π |~x − d~ez |3
Now using Griffiths Eq. (8.34), the electromagnetic angular momentum per unit volume
is
~`em = 0~x × (E ~ ×B ~ ) = µ0 qe qm
~x × (d~ × ~x )
(4π) r |~x − d~ez |3
2 3

µ0 dqe qm
= (~ez − ~er cos θ) .
(4π)2 r|~x − d~ez |3

The total angular momentum is L ~ em = R ~`em d3 x = R ~`em r2 dr dΩ. In Cartesian coor-

dinates, ~er =q~ex sin θ cos φ + ~ey sin θ sin φ + ~ez cos θ so that 02π ~er dφ = 2π~ez cos θ. Also,
R

|~x − d~ez | = r 1 − 2(d/r) cos θ + (d/r)2 . Combining ingredients gives

~ µ0 qe qm Z ∞
1
Lem = (~ez d) f (d/r) dr ,
8π 0 r2
where Z 1 (1 − x2 ) dx 4
f (a) = = min(1, a−3 ) for a > 0 .
−1 (1 − 2ax + a )2 3/2 3
This gives " #
~ µ0 qe qm 4 Z d r3 dr Z ∞ dr µ0
Lem = (~ez d) + = qe qm~ez .
8π 3 0 dr 3 2 d r 2 4π

Problem 2: Maxwell Stress Tensor

The Maxwell stress tensor in vacuum is
1 1 1
   
Tij ≡ 0 Ei Ej − E 2 δij + Bi Bj − B 2 δij .
2 µ0 2
For a linear medium, 0 and µ0 are replaced by  and µ, respectively.

1
a) The given E ~ implies cB ~ = ~ey E cos(kz − ωt). Substituting into the Maxwell
stress tensor formula and using 0 µ0 = 1/c2 gives Tij = 0 for all (i, j) except
Tzz = −0 E 2 cos2 (kz − ωt). Performing the time-average gives
1
hTzz i = − 0 E 2 .
2
Clearly, this is easily generated to a plane wave traveling in arbitrary direction
~n, by Tij = − 21 0 E 2 ni nj .
q
~ implies B
b) The given E ~ = E µ/2[−~ex sin(kz − ωt) + ~ey cos(kz − ωt)]. Again
the Maxwell stress tensor components all vanish except
1
Tzz = − E 2 .
2
This value is constant, unchanged by time-averaging.
c) For a constant electric field in the x-direction,
1 1
Txx = 0 E 2 , Tyy = Tzz = − 0 E 2 .
2 2
d) For a constant magnetic field in the y-direction,
1 2 1
Tyy = B , Txx = Tzz = − B2 .
2µ 2µ

e) For a constant magnetic field in the ~n-direction,

1 2
Tij = B (ni nj − 12 δij )
µ0
where ni are the Cartesian components of ~n. Defining (θ, φ) to be the direc-
tion in which ~n points, we have
nx = sin θ cos φ , ny = sin θ sin φ , nz = cos θ .
The angular average of ni nj is
1 Z 1 Z π Z 2π dφ
hni nj i ≡ ni nj dΩ = ni nj sin θ dθ .
4π 2 0 0 2π
The φ integration vanishes for all components except the diagonal ones, which
give
1
hni nj i = δij
3
and therefore
B2 1
Tij = −pmag δij , pmag = = umag .
6µ0 3

2
Problem 3: Griffiths Problem 9.5 (p. 373)
Let us denote the total wave by f (z, t). We have
gI (z − v1 t) + hR (z + v1 t) , z < 0 ,

f (z, t) =
gT (z − v2 t) , z>0.
The boundary condition [f ] = 0 at z = 0 (where the square brackets denote the jump in
the quantity evaluated across z = 0) becomes
gI (−v1 t) + hR (v1 t) = gT (−v2 t) (1)
while the boundary condition [∂f /∂z] = 0 gives
gI0 (−v1 t) + h0R (v1 t) = gT0 (−v2 t) , (2)
where a prime denotes differentiation with respect to the argument, i.e. g 0 (u) = dg/du.
Now, we use a trick, noting that
1 d
gI0 (−v1 t) = − gI (−v1 t)
v1 dt
and similarly for h0R (v1 t) and gT0 (−v2 t). (This follows for u = −v1 t because du = −v1 dt.)
Equation (2) therefore can be written
1 d 1 d 1 d
− gI (−v1 t) + hR (v1 t) = − gT (−v2 t) . (3)
v1 dt v1 dt v2 dt
This equation can be integrated to give
1 1 1
− gI (−v1 t) + hR (v1 t) = − gT (−v2 t) + C , (4)
v1 v1 v2
where C is an arbitrary constant. Combining equations (1) and (4), we obtain
v2 − v1 2v2
hR (v1 t) = gI (−v1 t) + C 0 , gT (−v2 t) = gI (−v1 t) + C 0 , (5)
v1 + v2 v1 + v2
where C 0 = Cv1 v2 /(v1 + v2 ).
Consider now the result for hR (v1 t). Let t = t0 + (z/v1 ). From equation (5) it follows
v2 − v1
hR (z + v1 t0 ) = gI (−z − v1 t0 ) + C 0 .
v1 + v2
To evaluate gT , let t = t0 − (z/v2 ). Substituting into the second of equations (5), we get
2v2 v1
 
gT (z − v2 t0 ) = gI (z − v2 t0 ) + C 0 .
v1 + v2 v2
Thus, up to an additive constant (which must be determined by initial conditions),
v2 − v1 2v2
hR (u) = gI (−u) , gT (u) = gI (uv1 /v2 ) .
v1 + v2 v1 + v2
This result can also be obtained using equations (9.25) and (9.30) of Griffiths by
representing an arbitrary function gI (u) by a Fourier integral (linear superposition of
waves).

3
Problem 4: Griffiths Problem 9.16 (p. 392)
In this problem the electric fields are all in the y-direction, and the directions of the
wavevectors are

k̂I = ~ex sin θI + ~ez cos θI , k̂R = ~ex sin θR − ~ez cos θR , k̂T = ~ex sin θT + ~ez cos θT .
~ = v −1 k̂ × E,
Using B ~ he magnetic field amplitude vectors are therefore

~I = 1
B Ẽ0I (~ez sin θI − ~ex cos θI ) ,
v1
~R 1
B = Ẽ0R (~ez sin θR + ~ex cos θR ) ,
v1
~T 1
B = Ẽ0T (~ez sin θT − ~ex cos θT ) .
v2
The first of Griffiths Eqs. (9.101) gives 0 = 0 but the other equations give nontrivial
boundary conditions:
1 1
(Ẽ0I sin θI + Ẽ0R sin θR ) = Ẽ0T sin θT ,
v1 v2
Ẽ0I + Ẽ0R = Ẽ0T ,
1 1
(−Ẽ0I cos θI + Ẽ0R cos θR ) = − Ẽ0T cos θT .
µ1 v1 µ2 v2
Using Snell’s law, the first equation is equivalent to the second one. Combining the
second and third equations gives
!
1 − αβ 2
Ẽ0R = Ẽ0I , Ẽ0T = Ẽ0I ,
1 + αβ 1 + αβ
where
cos θT µ1 v1 µ 1 n2
α≡ , β≡ = .
cos θI µ2 v2 µ 2 n1
In general Ẽ0R does not vanish. (If µ1 = µ2 = µ0 , then Ẽ0R vanishes if and only
if 1 = 2 , in which case there is no dielectric interface. Strictly speaking, Ẽ0R can
vanish if 1 6= 2 , but only if µ1 /µ2 differs from 1. This would be exceedingly unusual
because the permeability is nearly always very close to µ0 except for ferromagnets and
superconductors, but these materials are conductors not dielectrics.)
In the limit of normal incidence, θI = θT = 0, α = 1 and the results reduce to
Griffiths Eq. (9.109) for α = 1. The reflection and transmission coefficients are
2 !2 2 !2
E0R 1 − αβ 2 v 2 E0T cos θT 2
 
R= = , T = = αβ =1−R .
E0I 1 + αβ 1 v 1 E0I cos θI 1 + αβ

4
 1

0.5

0
0 20 40 60 80
Theta

-0.5

-1

Figure 1: Amplitudes of the reflected (red) and transmitted (green) electric field, relative
to the incident field, for n2 /n1 = 1.5. Theta is the angle of incidence, in degrees.

5
Problem 5: Griffiths Problem 9.19 (p. 396)
a) The skin depth is given by Griffiths Eqs. (9.126) and (9.128),
s s −1/2
2
2  σ

d= 1+ − 1 .
µω 2 ω

Taking the limit σ  ω and expanding, we get

s
2 
d≈ , σ  ω .
σ µ

For water, µ = µ0 , (/0 )1/2 = 1.33 is the index of refraction, and σ −1 = 2.5 ×
105 in SI units (Ohm-meter) from Griffiths Table 7.1, giving d = 1.8 × 103 m.
Please note, the static dielectric constant /0 = 80.1 is inappropriate, unless
the frequency of electromagnetic waves is well below the vibrational frequency
of the hydrogen bonds in the water. At wavelengths much shorter than about
one centimeter, but not so short as to excite electronic transitions in the
ultraviolet, the index of refraction of water is (/0 )1/2 = 1.33.
q
b) In the limit σ  ω, the skin depth becomes d ≈ 2/(σµω) from the formula
above. Since λ = 2π/k, k −1 = λ/(2π). Using σ −1 = 10−7 in SI units,  = 0 ,
µ = µ0 , and ω = 1015 rad/s, we find σ/(ω) ∼ 103 so the metal is a good
conductor, and d = 13 nm, compared with the free-space wavelength of light
of 1800 nm (for ω = 1015 rad/s; this is actually in the infrared). Thus, light
is damped before travelling even one wavelength into a conductor. This is
why metals are opaque.

c) From Griffiths Eqs. (9.134) and (9.136), the phase shift of the magnetic field
is φ = tan−1 (1/(kr d)). For a good conductor, kr d = 1 so that the phase
shift is π/4, with the magnetic field lagging the electric field. The ratio of
magnetic to electric field amplitudes is (Griffiths Eq. 9.137)
v s
u 2
B0 σ µσ
u  r
= µ 1 + ≈ .
t
E0 ω ω

Using the numbers given, this ratio is 1.12 × 10−7 s/m = 34/c.