1

Assignment 1 — Solutions 1/8/04

Problem 1.1

(a) If there is net charge inside the conductor, the electric field (being somehow proportional to charge) will be
nonzero there. A nonzero electric field means there is a net force on any charge in that region so the charges will move.
We know from experience that [non-super-] conductors at equilibrium don't have currents running around in them,
hence the charges must rearrange themselves in such a way that no more movement is possible, i.e. the electric field
and thus the charge density inside the conductor must vanish. Of course we still have a bunch of charges around, and
the only place they can go is to the surface.

(b) Consider two Gaussian surfaces S1 in the interior and S2 in the exterior of a closed
hollow conductor (shaded in the example). If the charges are all outside, S1 encloses no
charge and so by Gauss' law the total E -field flux through it is zero. Now choose S1 to
S2
be an infinitesimally small surface containing any point within the interior, so the flux is
approximately equal to the E -field at the point [times some constant]. Since the flux S1
vanishes, so must the E -field. On the other hand if all the charges are inside, choosing
S2 to enclose the conductor and applying Gauss' law shows that the flux through S2 is
not zero, i.e. there must be some electric field somewhere on the surface, thereby
proving the second claim.

(c) Consider an infinitesimally small patch of the surface of the conductor with area dA.
Choose a "pillbox" Gaussian surface S with end caps infinitesimally close to the exterior and
interior of the conductor surface, and sides perpendicular to the surface (example of a cross
section shown in the figure). In (a) we argued that there is no electric field inside the
conductor, so the flux through S must come solely from the "top" (the exterior end cap) and S σ
sides. The contribution from the sides is due to the component of the E -field parallel to the
surface (determined by the normals of the sides of S ), but if this is nonzero there will be a
net flow of charge along the surface, which cannot be. So the E -field must be perpendicular

E dA . The charge enclosed is approximately s dA . By Gauss' law we conclude E = s ê e0 .

to the surface and the only contribution to the flux is from the top, which is approximately

Problem 1.5

FHrL = ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ ÅÅÅÅÅÅÅÅÅÅÅÅÅÅ J1 + ÅÅÅÅÅÅÅÅÅ N = ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ J ÅÅÅÅÅÅÅÅÅÅÅÅÅÅ + ÅÅÅÅÅ e-a r N

q e-a r ar q e-a r a
4 p e0 r 2 4 p e0 r 2

Use equation (1.28) in Jackson, but be careful in calculating the Laplacian since “2 H1 ê rL = -4 p dHrL from equation
(1.31). It's easiest to do this in two steps:
 2

Ä É
q 1 ∑ ÅÅ i yÑÑ
r = -e0 “2 F = -e0 ÅÅÅÅ2ÅÅÅ ÅÅÅÅÅÅÅÅÅ jijr2 ÅÅÅÅÅÅÅÅÅÅ zyz = - ÅÅÅÅÅÅÅÅÅÅ ÅÅÅÅ2ÅÅÅ ÅÅÅÅÅÅÅÅÅ ÅÅÅÅr2 jj-a ÅÅÅÅÅÅÅÅÅÅÅÅÅÅ + e-a r ÅÅÅÅÅÅÅÅÅ ÅÅÅÅÅ - ÅÅÅÅÅÅÅÅÅ e-a r zzÑÑÑÑ
4 p r ∑r ÅÅÇ k {ÑÑÖ
1 ∑ ∑F e-a r ∑ 1 a2
r ∑r k ∑r {
Ä
Å É
Ñ
ÅÅ ÑÑy
∑r r
q 1 ∑ i
r 2

= - ÅÅÅÅÅÅÅÅÅÅ ÅÅÅÅ2ÅÅÅ ÅÅÅÅÅÅÅÅÅ jje-a r ÅÅÅÅ-a r + r2 ÅÅÅÅÅÅÅÅÅ ÅÅÅÅÅ - ÅÅÅÅÅÅÅÅÅ r2 ÑÑÑÑzz

ÅÇ 2 ÑÖ{
∑ 1 a
4 p r ∑r k
2

Ä É
q e-a r i ÅÅÅ a2 ÑÑÑ
∑r r
y
= - ÅÅÅÅÅÅÅÅÅÅ ÅÅÅÅÅÅÅÅ2ÅÅÅÅÅÅ jj-a ÅÅÅ-a r + r2 ÅÅÅÅÅÅÅÅÅ ÅÅÅÅÅ - ÅÅÅÅÅÅÅÅÅ r2 ÑÑÑ - a + ÅÅÅÅÅÅÅÅÅ r2 ÅÅÅÅÅÅÅÅÅ ÅÅÅÅÅ - a2 rzz
4 p r k ÅÅÇ 2 ÑÑÖ
∑ 1 ∑ ∑ 1
∑r r ∑r ∑r r {
q -a r ij a2 y i y
j ÅÅÅÅÅÅÅÅÅ + ÅÅÅÅÅÅÅÅÅ + “2 ÅÅÅÅÅ - ÅÅÅÅÅÅÅÅÅ zz = - ÅÅÅÅÅÅÅÅÅÅ e-a r jj ÅÅÅÅÅÅÅÅÅ - 4 p dHrLzz
a3 a a
k r r { k 2 {
2 3
1 q
= - ÅÅÅÅÅÅÅÅÅÅ e
4p 2 r 4p
q a -r a 3
= q dHrL - ÅÅÅÅÅÅÅÅÅÅÅÅÅ e
8p

In the last step I've taken the liberty to set e-a r Ø 1 in the term containing the d-function. We see that r is a point
charge q at the origin plus a spherically symmetric, exponentially decaying charge density of opposite sign (assuming
a > 0).

Problem 1.10

R = †x” - x” '§ = r and ∑ ê ∑n' = ∑ ê ∑r:

Use equation (1.36) with r = 0 and V a sphere of radius b centered about the origin, so that in spherical coordinates

ÅÄÅ 1 ∑F ∑ 1 ÉÑÑ 1 ÷”
FH0L = ÅÅÅÅÅÅÅÅÅÅ ® ÅÅÅÅ ÅÅÅÅÅ ÅÅÅÅÅÅÅÅÅÅ - F ÅÅÅÅÅÅÅÅÅ J ÅÅÅÅÅ NÑÑÑÑ a ' = ÅÅÅÅÅÅÅÅÅÅ ® ÅÅÅÅÅ E ÿ r` a' + ÅÅÅÅÅÅÅÅÅÅ ® ÅÅÅÅ2ÅÅÅ a '
4 p S ÅÇ r ∑ r ∑r r ÑÖ
1 1 1 F

÷”
4p S r 4p S r

= ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ ® E ÿ r` a' + ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ2ÅÅ ® F a'

1 1
4pb S 4pb S
÷” ÷” ÷”
The first term vanishes by application of the divergence theorem since òS E ÿ r` a' = ŸV “ ÿ E 3 x = ŸV Hr ê e0 L 3 x = 0.

is all it takes for this to hold for points not at the origin, i.e. we have proven FHx” L = XF\sphere centered at ÷x” for any x” .
The second term is the average of F over the surface S of the sphere as promised. Clearly a translation of coordinates

Problem 1.15

The system has n surfaces Si , 1 § i § n with charge density si and total charge Qi = ŸS si ai on the ith surface. For
convenience define the collection of all surfaces S ª ‹ni=1 Si . The electrostatic energy of the system is according to
i

(1.53):

÷”L FHx
÷”L 3 x = ÅÅÅÅÅ ÷” ÷”
W = ÅÅÅÅÅ ‡ rHx ‡ sHx L FHx L a
1 1
2 V 2 S

where r Ø s ª ⁄ni=1 si since the only charges around are those on the surfaces. Consider the change in energy due to a
change in fields F Ø F ' = F + dF and charge configuration s Ø s' = s + ds:

dW = WHs + ds, F + dFL - WHr, FL = ÅÅÅÅÅ ‡ Hds F + s dF + ds dFL a

1
2 S
 3

Applying Green's reciprocation theorem (problem 1.12) with no volume charges and F, F', s, s' as given above:

‡ s HF + dFL a = ‡ Hs + dsL F a ï ‡ s dF a = ‡ ds F a

dW = ‡ Jds F + ÅÅÅÅÅ ds dFN a

S S S S
1
S 2

Now suppose that F§Si = Fi = constant on each surface. ŸS ds F 3 x = ‚ Fi ŸS dsi 3 x = 0 since ŸS dsi 3 x is the
n

÷”
i=1

potential by sHx” L ê e0 = Hlim+ -lim- L E Hx” L ÿ n` = Hlim+ -lim- L ∑FHx” L ê ∑n (1.22) where n` is the unit surface normal and the
i i

(zero) total change in charge on the ith surface, so the first term vanishes. Surface charge density is related to the

rather schematic notation lim≤ means to take the limit approaching the point x” on the surface from above/below. So

s' = s + ds = e0 Hlim+ -lim- L ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ = e0 Hlim+ -lim- L

∑HF + dFL ∑ F'
ÅÅÅÅÅÅÅÅÅÅÅÅÅ
÷”
∑n ∑n
ds = e0 Hlim -lim L ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ = e0 Hlim+ -lim- L “ HdFL ÿ n`
+ - ∑HdFL

÷” ÷”
∑n
dW = ÅÅÅÅÅ ‡ ds dF a = ÅÅÅÅÅÅÅ ‡ dF Hlim+ -lim- L “ HdFL ÿ n` a = ÅÅÅÅÅÅÅ Hlim+ -lim- L ‡ dF “ HdFL ÿ n` a
1 e0 e0
2 S 2 S 2 S

The last is possible because potentials are continuous across surface charge distributions, so lim+ dF = dF = lim- dF
and we would be smart to do the integrals above/below the respective surfaces (where the integrands are well-behaved)

÷” ÷” ÷” ` ÷”
Ÿ “ ÿ A 3 x = ò A ÿ n a holds only for continuous vector fields A . There is another hitch —  in the theorem must be
and then take the limit approaching the surface, instead of vice versa. This is important because the divergence theorem

the integral over the ith surface Hlim+ -lim- L òS f a. If Si is closed then everything is fine, otherwise
a closed surface containing , but our S may be the union of a set of disjoint surfaces. To get around this let's consider

÷”
i

To get around this let's consider an integral over the ith open surface, Hlim+ -lim- L ŸS A ÿ n` a.
i

Define the closed surfaces Si+ /Si- with one boundary just above/below Si and such that Si+ and

÷” ÷” ÷”
Si +

Hlim+ -lim- L ŸS A ÿ n` a = lim+ òS+ A ÿ n` a - lim- òS- A ÿ n` a since the integrals over the extra
Si- share almost all of three other boundaries (purple in the example sketch). We can write
Si
Si -
i i i

boundaries either cancel or vanish in the limit. For a closed surface Si we can simply take Si+ /Si-
to be the closed surfaces infinitesimally close above/below Si . Then we can apply the divergence
theorem after all with Vi≤ being the volume enclosed by Si≤ :

÷” ÷” ÷”
Hlim+ -lim- L ‡ dF “ HdFL ÿ n` a = lim+ ® dF “ HdFL ÿ n` a - lim- ® dF “ HdFL ÿ n` a

÷” ÷” ÷” ÷”
Si Si+ Si-

= lim+ ‡ “ ÿ AdF “ HdFLE 3 x - lim- ‡ “ ÿ AdF “ HdFLE 3 x

÷” ÷” ÷”
Vi+ Vi-

‡ “ ÿ AdF “ HdFLE 3 x = ‡ °“ HdFL• 3 x + ‡ dF “2 HdFL 3 x

2

Vi≤ Vi≤ Vi≤

“2 HdFL is the change in volume charge distribution, because if there were some volume charge r (which does not
include the surface charges) we would have r + dr = e0 “2 HF + dFLïdr = e0 “2 HdFL. However dr = 0 everywhere,

÷” ÷”
so the second term vanishes and we're left with a manifestly nonnegative quantity. Since Vi+ contains Vi- , the quantity
lim+ ‡ °“ HdFL• 3 x - lim- ‡ °“ HdFL• 3 x is the integral of a nonnegative quantity over a volume Vi+ - Vi- and so
2 2
+ -
Vi Vi
must also be nonnegative. Thus after much ado
 4

÷” ÷”
dW = ÅÅÅÅÅÅÅ Hlim -lim L ‡ dF “ HdFL ÿ n a = ÅÅÅÅÅÅÅ „ Hlim -lim L ‡ dF “ HdFL ÿ n a
n
e0 + - ` e0 + - `
2 S 2 Si
i=1

ji + ÷” ÷” 2 3 zy ÷”
= ÅÅÅÅÅÅÅ „ jjlim ‡ °“ HdFL•  x - lim ‡ °“ HdFL•  xzz = ÅÅÅÅÅÅÅ ‚ ‡ °“ HdFL• 3 x
n
e0 e0 n
k {
2 3 - 2

2 Vi
+ Vi
- 2 i=1 Vi
+ -V -
i
i=1
¥ 0

Hence this equipotential configuration F is a minimum-energy configuration, as was to be shown. Note that this is a
proof of F being a global minimum because dF was completely arbitrary, not necessarily a small variation from F.

Problem 1.16

Like in 1.15, let the system have n surfaces Si bounding volumes Vi , 1 § i § n with
charge density si and total charge Qi = òS si ai on the ith surface. The collection
of surfaces S ª ‹ni=1 Si bounds an external volume V + V0 (shaded in the example).
V1
i

÷”
Call the potential F. Since the surfaces are conductors F§Si = Fi = constant on each V
surface, and si = e0 “ Fi ÿ n` i where n` i is the outward normal of the Si (you proved V2
V0
this in terms of the E -field in problem 1.1c). Keeping in mind that the electric field Vn
vanishes inside all the conductors, the system has electrostatic energy

÷” ÷” ÷” ÷” ÷” ÷”
W = ÅÅÅÅÅÅÅ ‡ “ F ÿ “ F 3 x = ÅÅÅÅÅÅÅ ‡ “ F ÿ “ F 3 x + ÅÅÅÅÅÅÅ ‡ “ F ÿ “ F 3 x
e0 e0 e0
2 all space 2 V 2 V0

After the introduction of an uncharged insulator with surface S0 bounding volume V0 , the potential and charge densi-

÷ ”£ ÷ ”
ties change to F£ = F + dF , Fi£ = Fi + dFi , and si£ = si + dsi (1 § i § n). The total charge remains fixed so
òS dsi a = 0, and the new conductor is at equipotential F£ §S0 = F£0 = constant with E = “ F£ = 0 inside [V0 ]. The
i

electrostatic energy is now

÷” ÷”
W £ = ÅÅÅÅÅÅÅ ‡ “ HF + dFL ÿ “ HF + dFL 3 x
e0

÷” ÷” ÷” ÷” 3 ÷” 2 3
2 V

dW = W - W = ÅÅÅÅÅÅÅ ‡ “ dF ÿ “ dF  x + e0 ‡ “ dF ÿ “ F  x - ÅÅÅÅÅÅÅ ‡ °“ F•  x
£ e0 3 e0
2 V V 2 V0

As in the previous problem, “2 HdFL is the change in volume charge distribution which is zero in V (note that this is
bounded by S + S0 ). The conductors must still be equipotentials, so Fi£ = constant, which means dFi = Fi - Fi£ is also

÷” ÷” ÷”
si£ = e0 “ Fi£ ÿ n` i = e0 “ HFi + dFi L ÿ n` i , so dsi = e0 “ dFi ÿ n` i for 1 § i § n. Thus
constant (being the difference of two constant quantities). The new charge distributions must still also satisfy
 5

÷” ÷” ij÷ ” ÷” yz 3 ÷” ÷”
‡ “ dF ÿ “ dF  x = ‡ jjj“ ÿ IdF “ dFM - dF “ ¨¨¨¨Æzzz  x = ‡ “ ÿ IdF “ dFM  x
V k 0 {
´¨¨¨¨¨¨≠dF
3 2 3
V V

÷” ÷”
= ® dF “ dF ÿ n` a = „ dFi ® dsi a + ® HF0 - F£0 L “ dF0 ÿ n` a
n

S+S0 ´¨¨¨¨¨¨¨¨
Si ¨¨¨≠¨¨¨¨¨¨¨¨¨Æ S0

÷” ÷”
i=1
0

= - ® F0 “ dF0 ÿ n` a - F£0 ® “ dF0 ÿ n` a

S0 ´¨¨¨¨¨¨¨¨
S0 ¨¨¨¨¨¨¨¨¨ ≠¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨Æ

÷” ÷” 2 3
0

® F0 “ dF0 ÿ n a = - ‡ °“ dF•  x § 0
`
S0 V

Similarly

÷” ÷” 3 ij÷ ” ÷” yz 3 ÷” ÷”
‡ “ dF ÿ “ F  x = ‡ jjj“ ÿ IF “ dFM - F “ ¨¨¨¨Æzzz  x = ‡ “ ÿ IF “ dFM  x
V k 0 {
´¨¨¨¨¨¨≠dF
2 3
V V

÷” ÷”
= ® F “ dF ÿ n` a = „ Fi ® dsi a + ® F0 “ dF0 ÿ n` a
n

S+S0 ´¨¨¨¨¨¨¨¨
Si ¨¨¨≠¨¨¨¨¨¨¨¨¨Æ S0

÷”
i=1
0

= ® F0 “ dF0 ÿ n a § 0
`
S0

We have shown that the first and second terms of dW are nonpositive, and the third one is already manifestly so.
Hence dW § 0, i.e. introduction of the new conductor lowers the electrostatic energy.