Jackson 3.3
Jackson 3.3
Jackson 3.3
3
Stanislav Boldyrev, Physics Department, UW-Madison
The (correct!) answer is given in Jackson. The problem is, however, traditionally considered difficult by
students and teachers. Here we present its full solution.
Jackson 3.3: A thin, flat, conducting, circular disc of radius R is located in the x − y plane with its center
at the origin, and is maintained at a fixed potential V . With the information that the charge density on a
disc at fixed potential is proportional to (R2 − ρ2 )−1/2 , where ρ is the distance out from the center of the
disc,
a) show that for r > R the potential is
∞ ( )2n
2V R ∑ (−1)n R
Φ(r, θ, ϕ) = P2n (cos θ);
π r n=0 2n + 1 r
Solution:
The answer is given by formulae (26, 27, 28) below. There are several ways to solve this problem. One way
is to expand the potential and the charge density in the Legendre polynomials and to solve the resulting
equations for the coefficients of expansion. Here we use a different method, based on the Green function.
We write the Green function (assuming that the potential is zero at r → ∞) in spherical coordinates as:
∞ ∑
∑ l l
1 1 r<
G(x, x′ ) = − ′
= − Y ∗ (θ′ , ϕ′ )Ylm (θ, ϕ), (1)
4π|x − x | 2l + 1 r> lm
l+1
l=0 m=−l
where r< is the smaller of the two parameters r = |x| and r′ = |x′ |, and r> is the larger. The charge density
concentrated on the disc is represented in spherical coordinates by
Q θ(R − r)
ρ(x)d3 x = δ(cos θ) √ d cos θ dϕ r2 dr, (2)
2πR r R2 − r 2
where θ(R − r) is a Heaviside step function, and Q is the charge of the disc. One can easily verify that
∫
ρ(x)d3 x = Q. (3)
With the aid of the Green function the potential is found in the whole space as:
∫
1
Φ(x) = − G(x, x′ )ρ(x′ )d3 x′ . (4)
ϵ0
Since only the m = 0 terms survive, we get:
∞
Q ∑
Φ(r, θ) = Il (r)Pl (0)Pl (cos θ), (5)
4πϵ0 R
l=0
1
and since only the even-order Legendre polynomials are nonzero at cos θ = 0, we get:
∞
Q ∑
Φ(r, θ) = I2n (r)P2n (0)P2n (cos θ). (6)
4πϵ0 R n=0
In these expressions the integral Il (r) has the form for r < R:
∫R ∫r ∫R
r′ dr′ r′ dr′ rl dr′
l l+1
r<
Il (r) = l+1
√ ≡ √ + √ , (7)
r> R2 − r ′ 2 rl+1 R2 − r′ 2 r
r ′ l R2 − r ′ 2
0 0
These integrals can be especially easily calculated for the limit of vanishing r:
∫R
dr′ π
I2n (r)|r→0 = √ = δn,0 , (9)
R2 − r ′ 2 2
0
which means that the only nonvanishing integral corresponds to n = 0. The other simple case is r > R:
∫R ( )2n+1 ∫1 ( )2n+1
r′ dr′
2n+1
R 1 tn dt R (2n)!!
I2n (r)|r≥R = √ = √ = , (10)
r2n+1 R2 − r ′2 r 2 1−t r (2n + 1)!!
0 0
where the integral is done by effecting the change of variables t = (r′ /R)2 , and integrating by parts n times.
The limiting case (9) allows us to evaluate the potential at the center of the disc,
π Q
Φ0 = , (11)
2 4πϵ0 R
and since Φ = V at the disc we can express the potential everywhere in space as:
∞
2V ∑
Φ(r, θ) = I2n (r)P2n (0)P2n (cos θ). (12)
π n=0
Formula (12) provides the general solution of the problem, where the functions I2n (r) are given by (7, 8).
We now discuss in more detail the solution for r > R and r < R. The first case is relatively simple. The
second case often causes confusion, and it will be discussed in more detail.
The region r > R. Substituting in (12) expression (10), and noting that1
2
which coincides with Jackson’s solution.
The region r < R. In this case the functions I2n (r) are not powers of r, and the structure of the solution
is not that transparent. We can however expand these functions in the powers of r, by using the following
expansion in y = r/R < 1:
∑ ∞
1 (2m − 1)!! 2m
√ = y . (15)
1−y 2
m=0
(2m)!!
Substituting this expansion inside the integrals in (7) and integrating each term we get:
∑∞ ∞
∑ ∞
∑
(2m − 1)!! y 2m+1 (2m − 1)!! y 2m+1 (2m − 1)!! y 2n
I2n (r) = − + . (16)
m=0
(2m)!! (2m + 2n + 2) m=0 (2m)!! (2m − 2n + 1) m=0 (2m)!! (2m − 2n + 1)
Here the first term comes from the first integral in (7), and the last two terms from the second integral. The
last sum in (16) can be done in a closed form. For that we notice that this sum can be represented through
the integral in a complex z-plane as follows:
∑∞ I
(2m − 1)!! y 2n y 2n dz
= √ √ , (17)
m=0
(2m)!! (2m − 2n + 1) 4 z z 1−z
n
C
where the integration contour encircles the cut made along the interval (0, 1) as shown in Fig. (1). We leave
it to the reader to verify this fact.
Ԗ
0 1
Figure 1: Contour of integration C in Eq. (17). Here ϵ is a small parameter, and the limit ϵ → 0 should be
taken in the answer.
meaning that the sum in (17) is identically zero except for the case of n = 0. Combining the remaining two
sums, we then simplify:
∑∞
(2m − 1)!! (−1)(4n + 1) ( r )2m+1 π
I2n (r) = + δn,0 . (19)
m=0
(2m)!! (2m + 2n + 2)(2m − 2n + 1) R 2
We now consider the obtained solution separately in the half intervals 0 ≤ cos θ ≤ 1 and −1 ≤ cos θ ≤ 0.
In the whole interval [−1, 1] the Legendre polynomials form a complete and mutually orthogonal system
of functions in L2 space, so the expansion of our solution in Legendre polynomials (12) is unique. In each
of the half intervals, however, the system of even Legendre polynomials and the system of odd Legendre
polynomials independently provide complete sets of mutually orthogonal functions. A function defined in
a given half interval can be expanded there in either odd or even Legendre polynomials. Our result (12)
provides the expansion of the solution in the even Legendre polynomials. We will demonstrate now that in
each of the half intervals the expansion of the solution in the odd Legendre polynomials has a much simpler
form.
3
For that we need to know how to expand the odd Legendre polynomials in the even Legendre polynomials
in the interval [0, 1]. The answer is [see e.g., Byerly, 1959, An Elementary Treatise on Fourier’s Series and
Spherical, Cylindrical, and Ellipsoidal Harmonics, page 173]:
∫1
(−1)m+n (2n − 1)!!(2m + 1)!!
P2m+1 (µ)P2n (µ)dµ = . (20)
(2n)!!(2m)!!(2n + 2m + 2)(2m − 2n + 1)
0
∫1
1
Pl (µ)Pl (µ)dµ = , (21)
2l + 1
0
where
(−1)m+n (2n − 1)!!(2m + 1)!!(4n + 1)
Cm,n = . (23)
(2n)!!(2m)!!(2n + 2m + 2)(2m − 2n + 1)
Combining (12), (19), (22), and (23), we see that in the region 0 ≤ cos θ ≤ 1 and r < R (that is, above the
disc) our solution (12) can be rewritten in a very simple form:
2V ∑ (−1)m+1 ( r )2m+1
∞
Φ(r, θ) = V + P2m+1 (cos θ), (24)
π m=0 2m + 1 R
and, analogously, in the region −1 ≤ cos θ ≤ 0 and r < R (below the disc) we get (inverting the sign of the
odd Legendre polynomials):
2V ∑ (−1)m+1 ( r )2m+1
∞
Φ(r, θ) = V − P2m+1 (cos θ). (25)
π m=0 2m + 1 R
Note that the potential is equal to V at the disc (cos θ = 0), and the derivative of the potential normal to
the disc is discontinuous across the disc, which, as the reader may check, is consistent with the discontinuity
of the normal component of the electric field due to the disc charge density.
The answer: Formula (12) or its simplifications in different regions (14), (24), and (25) constitute the
complete solution of the problem. We summarize it here:
∞ ( )2n+1
2V ∑ (−1)n R
Φ(r, θ) = P2n (cos θ), r > R, 0 ≤ θ ≤ π, (26)
π n=0 2n + 1 r
2V ∑ (−1)m+1 ( r )2m+1
∞
Φ(r, θ) = V + P2m+1 (cos θ), r < R, 0 ≤ θ ≤ π/2, (27)
π m=0 2m + 1 R
2V ∑ (−1)m+1 ( r )2m+1
∞
Φ(r, θ) = V − P2m+1 (cos θ), r < R, π/2 ≤ θ ≤ π. (28)
π m=0 2m + 1 R
4
It may seem that the inner solution (27, 28) expanded in the odd Legendre polynomials cannot match the
outer solution (26) expanded in the even Legendre polynomials at r = R, which sometimes leads to the
(incorrect) conclusion that the answer given in Jackson is wrong. The confusion stems from the fact that for
r < R, the solution is not expanded in the odd Legendre polynomials in the whole interval [−1, 1]. Rather,
it is expanded in this interval in the functions P2m+1 (cos θ)sign(cos θ), which are not orthogonal to the even
Legendre polynomials. As unobvious as it may seem, the inner solution (27)-(28) does match the outer
solution (26) at r = R; this can be verified by using Eqs. (22) and (23).
It is important to note that it would be incorrect to look for the solution in the inner regions r < R (both
above and below the disc) in the form similar to Eq. (26),
∑ ( r )2n
Φ(r, θ) = C2n P2n (cos θ). (29)
n
R
Indeed, such an expansion would hold for a function satisfying the Laplace equation, while the potential in
the inner regions does not satisfy the Laplace equation. Rather, it satisfies the Poisson equation with the
charge density (2) that expansion in the even Legendre polynomials is nonzero in these regions. Similarly,
it would be incorrect to look for the solution in the outer region (r > R, 0 < θ < π) in the form
∑ ( )2n+2
R
Φ(r, θ) = C2n+1 P2n+1 (cos θ)sign(cos θ), (30)
n
r
since the potential must satisfy the Laplace equation in the outer region, while these expansion functions,
obviously, do not.
One needs to use the even Legendre polynomials for expanding the Green function, as the source δ(cos θ) is
expandable in the even Legendre polynomials while it is not expandable in the odd ones. Indeed, the delta
function does not belong to L2 , and it should not necessarily be expandable in a set of functions complete
in L2 (0, 1). Alternatively, when one solves the problem without using the Green function, but rather by
expanding the solutions in orthogonal functions and matching them at the boundary, it is convenient to use
the odd Legendre polynomials for r < R, as each of them automatically satisfies the boundary condition
P2m+1 = const = 0 at the disc.
The capacitance of the disc. According to formula (11), the disc capacitance is