Image Charges for Two Concentric Grounded Spheres—C.E.

Mungan, Spring 2020

First consider the problem of finding the image charge Q1 at radius R1 inside a grounded
metal sphere of radius R due to a point charge q located at radius r outside the sphere, as
sketched in Fig. 1.

R r2
r1
B θ R1 A
Q1 r q

Figure 1
Clearly Q1 must be off center. (We could add a second image charge at the center of the
sphere if we wanted the potential of the sphere to be nonzero.) By symmetry it must be on the
line between the center of the sphere and charge q, and it must be of opposite sign as q. The
quickest way to find the values of Q1 and R1 are to zero out the potential at two points on the
surface of the sphere. At point A we must have
kQ1 kq
+ =0 (1)
R − R1 r − R
where k is the Coulomb constant, whereas at point B we require
kQ1 kq
+ = 0. (2)
R + R1 r + R
Simultaneous solution of Eqs. (1) and (2) leads to
R2
R1 = (3)
r
and
R
Q1 = −q . (4)
r
Note that R1 < R and Q1 < q . To verify that this solution works for all other points on the
sphere, consider the arbitrary asterisked point at angle θ in Fig. 1. From the law of cosines we
have

1
 ⎛ R 2 ⎞2 R2
r1 = ⎜⎜ ⎟⎟ + R 2 − 2R cosθ (5)
⎝ r ⎠ r

using Eq. (3), and likewise

r ⎛ R 2 ⎞2 R2 r
r2 = r 2 + R 2 − 2Rr cosθ = R 2 + ⎜⎜ ⎟⎟ − 2R cosθ = r1 . (6)
R ⎝ r ⎠ r R

Substituting Eqs. (4) and (6) into the following equality verifies that
kQ1 kq
+ =0 (7)
r1 r2

as desired.
This idea works equally well for a point charge q at radius r inside a grounded metal sphere
of radius R. Adding primes to the left-hand sides of Eqs. (3) and (4) to indicate the charge and
position of this external image charge, we thus have
R2
R1ʹ = (8)
r
and
R
Q1ʹ = −q . (9)
r
Observe that R1ʹ > R and Q1ʹ > q , in contrast to the two inequalities for the inner image charge
given in the line following Eq. (4). We are going to be interested in image charges outside a
sphere of radius 3R, in which case Eqs. (8) and (9) become
9R 2
R1ʹ = (10)
r
and
3R
Q1ʹ = −q . (11)
r
Specifically, the problem by Seth Rittenhouse is to find the electric field at the asterisked
point in Fig. 2.

2
 3R
R
E=? q
2R 2R

Figure 2
Charge q induces image charge Q1 at R1 where we put r = 2R in Eqs. (3) and (4) to get
R1 = 12 R and Q1 = − 12 q . (12)

In addition, it induces image charge Q1ʹ at R1ʹ where we put r = 2R in Eqs. (10) and (11) to get
R1ʹ = 92 R and Q1ʹ = − 23 q . (13)

However, image charge Q1ʹ induces image charge Q2 at R2 where we put r = R1ʹ and q = Q1ʹ in
Eqs. (3) and (4) to get
R2 = 92 R and Q2 = + 13 q . (14)

Likewise, image charge Q1 induces image charge Q2ʹ at R2ʹ where we put r = R1 and q = Q1 in
Eqs. (10) and (11) to get
R2ʹ = 18R and Q2ʹ = +3q . (15)

These two image charges then induce the image charges

1 R and Q = − 1 q .
R3 = 18 (16)
3 6

and
R3ʹ = 81 R and Q3ʹ = − 92 q . (17)
2

It is clear that we are getting an even and an odd infinite series for both the inner and outer image
charges, given by
R2n+1 = 1 R9−n and Q2n+1 = − 1 q3−n (18)
2 2

and
R2n+2 = 2 R9−n and Q2n+2 = + 1 q3−n (19)
9 3

on the inside (where n = 0,1,2,... ) and by

3
 ʹ = 9 R9 n
R2n+1 ʹ = − 3 q3n
and Q2n+1 (20)
2 2

and
R2n+2
ʹ = 18R9 n and Q2n+2
ʹ = +3q3n (21)

on the outside. Finally the total electric field at the asterisked point in Fig. 2 is the sum of that
due to these four infinite series plus that due to the original point charge q,
⎡ ⎤
kq
∞⎢ −k 12 q3−n k 13 q3−n −k 23 q3n n
k3q3 ⎥
E= + ∑⎢ + + + (22)
2⎥
(4R)2 n=0 ⎢ 2R + 1 R9−n 2 2 2
⎣ ( 2 ) ( 2R + 92 R9 −n
) ( 9
2R + 2 R9 n
) ( )
n ⎥
2R +18R9 ⎦

directed leftward. To evaluate this expression as a multiple of kq / R 2 , copy and paste the
following line into Mathematica,
N[1/16+Sum[-(1/2/(3^n*(2+1/2/9^n)^2))+1/3/(3^n*(2+2/9/9^n)^2)-
((3/2)*3^n)/(2+(9/2)*9^n)^2+(3*3^n)/(2+18*9^n)^2,{n,0,Infinity}]]
to get a result of 0.000 526 476. After pasting it into Mathematica, you can convert this line to
“StandardForm” to get it to look more like Eq. (22). It makes sense that this answer is small,
slightly less than 1% of kq / (4R)2 , because the asterisked point lies in free space between two
grounded surfaces 180° away from charge q.
We can calculate a few other things. To find the potential at the asterisked point, simply
delete the squares in the five denominators in Eq. (22) to get V = 0.000 602 767 kq / R . To find
the total induced charge on the surface of the inner sphere, sum up all the image charges inside it
to obtain
∞ ⎛
1 1⎞ q 1 q
Qinner = ∑ ⎜ − ⎟ q3−n = − =− . (23)
n=0 ⎝
3 2⎠ 6 1−1/ 3 4

Since every field line that starts on the +q point charge must end on the two grounded spheres,
the total induced charge on the the surface of the outer sphere must be Qouter = −3q / 4 . If we try
to add the image charges outside it, however, we find a divergent series. Suppose we ignore that
fact and proceed nevertheless as in Eq. (23). We get
∞ ⎛
3⎞ 3q 1 3q
Qouter = ∑ ⎜3− ⎟ q3n = =− (24)
n=0 ⎝
2⎠ 2 1− 3 4

which is the right answer! That shows one can do sneaky things with divergent series. But there
is another way to get the answer which does not rely on such ruses. We can calculate the
potential at the origin in Fig. 2 due to the image charges for the outer sphere. That answer must
equal kQouter / (3R) since every point on its surface is a distance 3R away from the origin. Thus
⎡ n ⎤
kQouter ∞ ⎢ −k 23 q3 k3q3n ⎥ −kq ∞ −n −kq 1
=∑ 3 =
n ⎥ 6R ∑
+ = (25)
3R ⎢ 9 R9 n 18R9 6R 1−1/ 3
n=0 ⎣ 2 ⎦ n=0

4
which simplifies to Qouter = −3q / 4 . If you try this idea for the inner sphere, you will find that it
now gives a divergent series, but if you assume the same sum for 3n as in Eq. (24), namely −1/ 2 ,
you again get the right answer for Qinner.