Physics 505 Electricity and Magnetism Fall 2003

Prof. G. Raithel

Problem Set 8
Maximal score: 25 Points

1. Jackson, Problem 5.10 9 Points

a): In cylindrical coordinates, the 3D current density of a loop current I with radius a in the plane z = 0
centered at the origin is

j(ρ0 , z 0 , φ0 ) = φ̂0 Jφ (ρ0 , z 0 )

with Jφ (ρ0 , z 0 ) = Iδ(r0 − a)δ(z 0 ) .

R
(This applies because I = j(ρ0 , z 0 , φ0 ) · da0 , the integral taken over a plane of constant φ0 .)
1
Thus, using the expansion of |x−x0 | in Eq. 3.149 of Jackson, it is

Z
µ0 φ̂0 Jφ (ρ0 , z 0 ) 3 0
A(ρ, z, φ = 0) = d x
4π |x − x0 |
Z
µ0 (φ̂ · φ̂0 )Jφ (ρ0 , z 0 ) 3 0
Aφ (ρ, z) = d x
4π |x − x0 |
Z ∞ Z
µ0 I
Aφ (ρ, z) = cos φ0 δ(r0 − a)δ(z 0 ) cos(k(z − z 0 ))
π 2 k=0 ρ0 ,z0 ,φ0
" ∞
#
1 X
× I0 (kρ< )K0 (kρ> ) + {cos(m(φ − φ ))Im (kρ< )Km (kρ> )} ρ0 dρ0 dφ0 dz 0 dk
0
use φ = 0
2 m=1
Z
µ0 Ia ∞
Aφ (ρ, z) = cos(kz)I1 (kρ< )K1 (kρ> )dk q.e.d.
π k=0

1
b): Using the expansion of |x−x0 | in Problem 3.16b of Jackson, it is

Z ∞ ∞
X Z
µ0 I
Aφ (ρ, z) = cos φ0 δ(r0 − a)δ(z 0 ) exp(−k(z> − z< ))
4π k=0 m=−∞ ρ0 ,z 0 ,φ0

× exp(im(φ − φ0 ))Jm (kρ)Jm (kρ0 )ρ0 dρ0 dφ0 dz 0 dk use φ = 0

Z
µ0 Ia ∞
Aφ (ρ, z) = exp(−k(z> − z< ))J1 (kρ)J1 (ka)dk × 2
4
Zk=0
µ0 Ia ∞
Aφ (ρ, z) = exp(−k |z|)J1 (kρ)J1 (ka)dk q.e.d.
2 k=0

1
 1
Not required, but good exercise: The utilized expansion of Gfree (x, x0 ) = |x−x0 | is obtained as follows.
Using completeness relations for δ-functions, it is

4π
∆G = −4πδ(x − x0 ) = − δ(ρ − ρ0 )δ(z − z 0 )δ(φ − φ0 )
ρ
Z ∞ ∞
X
∆G = −2 dk kJm (kρ0 ) exp(−imφ0 )δ(z − z 0 )Jm (kρ) exp(imφ) (1)
k=0 m=−∞

Also, expanding the Green’s function it is

· ¸Z ∞ ∞
X
1 1
∆G = ∂ρ ρ∂ρ + 2 ∂φ2 + ∂z2 dk Ak,m (z|ρ0 , z 0 , φ0 )Jm (kρ) exp(imφ)
ρ ρ k=0 m=−∞
Z ∞ X∞ ½· 2 2
¸ ¾
d m 0 0 0
= dk − 2 Ak,m (z|ρ , z , φ ) Jm (kρ) exp(imφ)
k=0 m=−∞
dz 2 ρ
Z ∞ X∞ ½ ¾
1
+ dk ∂ρ ρ∂ρ Jm (kρ) Ak,m (z|ρ0 , z 0 , φ0 ) exp(imφ) by Bessel equation
k=0 m=−∞
ρ
Z ∞ X∞ ½· ¸ ¾
d2 2 0 0 0
= dk − k Ak,m (z|ρ , z , φ ) Jm (kρ) exp(imφ) (2)
k=0 m=−∞
dz 2

Equating the terms in Eqs. 1 and 2 in front of the orthogonal functions Jm (kρ) exp(imφ), and defining

Ak,m (z|ρ0 , z 0 , φ0 )
gk (z, z 0 ) =
−2kJm (kρ0 ) exp(−imφ0 )

the equation for the reduced Green’s function gk (z, z 0 ) is

· ¸
d2
− k gk (z, z 0 ) = δ(z − z 0 )
2
dz 2

To avoid divergence, the solution must be of the form

gk (z, z 0 ) = C exp(kz< ) exp(−kz> ) .

Inserting into that result into the differential equation for the reduced Green’s function and integrating over
an infinitesimal region that includes the δ-inhomogeneity, it is found

1
C=− .
2k

Inserting the results in reverse order, it is

2
 1
gk (z, z 0 ) = − exp(−k(z> − z< ))
2k
Ak,m (z|ρ0 , z 0 , φ0 ) = Jm (kρ0 ) exp(−imφ0 ) exp(−k(z> − z< ))
Z ∞ X∞
1
Gfree (x, x0 ) = = dk Jm (kρ0 )Jm (kρ) exp(−k(z> − z< )) exp(im(φ − φ0 )) q.e.d.
|x − x0 | k=0 m=−∞

c): case a):

B(ρ, z) = ∇ × A = (here) ∇ × (φ̂Aφ (ρ, z))

· ¸
1
B(ρ, z) = ρ̂ [−∂z Aφ (ρ, z)] + ẑ ∂ρ ρAφ (ρ, z)
ρ
· Z ∞ Z ∞ ½ ¾ ¸
µ0 Ia 1
B(ρ, z) = ρ̂ k sin(kz)I1 (kρ< )K1 (kρ> )dk + ẑ cos(kz) ∂ρ ρI1 (kρ< )K1 (kρ> ) dk
π k=0 k=0 ρ

On the z-axis, it is I1 (kρ< )K1 (kρ> ) = I1 (0)K1 (kρ) = 0 and thus Bρ = 0.

Also, using L’Hopital’s rule, it is for ρ → 0

· ¸
1 1 0
∂ρ ρI1 (kρ)K1 (ka) = K1 (ka) I1 (kρ) + kI1 (kρ)
ρ ρ
1
= 2kK1 (ka)I10 (kρ) = 2kK1 (ka) [I0 (kρ) + I2 (kρ)]
2
= kK1 (ka)I0 (kρ)
= kK1 (ka) for ρ → 0

and thus, using an integral table or Mathematica or equivalent, it is verified that

Z ∞
µ0 Ia µ0 Ia2
Bz = k cos(kz)K1 (ka)dk = √ 3
π k=0 2 z 2 + a2

µ0 Ia2
B(ρ = 0, z) = ẑ √ 3
2 z 2 + a2

case b):

· Z ∞ Z ∞ ½ ¾ ¸
µ0 Ia 1
B(ρ, z) = ρ̂ Sign(z) k exp(−k |z|)J1 (kρ)J1 (ka)dk + ẑ exp(−k |z|)J1 (ka) ∂ρ ρJ1 (kρ) dk
2 k=0 k=0 ρ

3
On the z-axis, it is Bρ = 0 and, taking the limit ρ → 0,

· ¸
1 1
∂ρ ρJ1 (kρ) = J1 (kρ) + kJ10 (kρ)
ρ ρ
1
= 2kJ10 (kρ) = 2k [J0 (kρ) + J2 (kρ)]
2
= k for ρ → 0

and thus, using an integral table or Mathematica or equivalent, it is verified that

Z ∞
µ0 Ia µ0 Ia a
Bz = exp(−k |z|)J1 (ka)kdk = √ 3
2 k=0 2 z + a2
2

µ0 Ia2
B(ρ = 0, z) = ẑ √ 3
2 z 2 + a2

4
2. Jackson, Problem 5.15 8 Points

Consider first a single wire with current I = Iẑ along the z-axis. Through variable separation of ∆ΦM = 0
in cylindrical coordinates and subsequent consideration of the ν = 0 terms it is seen that the magnetic
Iφ
potential is ΦM = − 2π (see Section 2.11 of Jackson). The validity of this result is verified by noting that
the correct H-field follows:

I
H = −∇ΦM = φ̂
2πρ

Note that the x-axis or another plane of constant φ needs to be “cut out” of the volume of interest.

f1-f2

f2 f f1 x
I2 I1
d/2 d/2

Figure 1: Geometry of the problem.

Now, consider two wires parallel to the z-axis, one with current I1 = Iẑ at (d/2, 0) and one with current
I2 = −Iẑ at (−d/2, 0). Then, by superposition it is found that

I
ΦM = (φ2 − φ1 )
2π

where the φi describe the azimuthal angles of the observation point with respect to the respective currents
2
Ii . Simple geometry shows that for ρ À d it is φ2 − φ1 = − d sin
ρ
φ
+ O( dρ2 ), where ρ and φ are the coordinates
of the observation point.

Thus

Id sin φ d2
ΦM = − + O( 2 ) , q.e.d.
2πρ ρ

Note that in the limit ρ À d the magnetostatic potential is valid without restriction, because the currents
through the volume of interest are limited to a small region in the center and add up to zero.

5
b): Through variable separation of the Laplace equation in cylindrical coordinates (2D) it is seen that the
potentials in the regions 1, 2 and 3 can be expanded as follows:

X∞
Id −1
Φ1 = − ρ sin φ + An ρn sin(nφ)
2π n=1
∞
X £ ¤
Φ2 = Bn ρn + Cn ρ−n sin(nφ)
n=1
X∞
£ ¤
Φ3 = Dn ρ−n sin(nφ)
n=1

The boundary conditions on the interfaces are, due to the absence of free currents,

1−2 2−3
n̂ · B1 |ρ=a = n̂ · B2 |ρ=a n̂ · B2 |ρ=b = n̂ · B3 |ρ=b
n̂ × H1 |ρ=a = n̂ × H2 |ρ=a n̂ × H2 |ρ=b = n̂ × H3 |ρ=b

In the given geometry and expressed with the magnetostatic potential, they are

1−2 2−3
µ0 ∂ρ Φ1 |ρ=a = µ∂ρ Φ2 |ρ=a µ∂ρ Φ2 |ρ=b = µ0 ∂ρ Φ3 |ρ=b
µ0 a1 ∂φ Φ1 |ρ=a = µ 1b ∂φ Φ2 |ρ=a µ a1 ∂φ Φ2 |ρ=b = µ0 1b ∂φ Φ3 |ρ=b

The resultant equations are

∞ ·
X ¸ ∞
X
Id −2 n−1
£ ¤
µ0 a δn,1 + µ0 An na sin(nφ) = µBn nan−1 − µCn na−n−1 sin(nφ)
n=1
2π n=1
X∞ · ¸ X∞
Id £ ¤
− a−2 δn,1 + An nan−1 sin(nφ) = Bn nan−1 + Cn na−n−1 sin(nφ)
n=1
2π n=1
∞
X X∞
£ ¤ £ ¤
µBn nbn−1 − µCn nb−n−1 sin(nφ) = −µ0 Dn nb−n−1 sin(nφ)
n=1 n=1
∞
X X∞
£ ¤ £ ¤
Bn nbn−1 + Cn nb−n−1 sin(nφ) = Dn nb−n−1 sin(nφ)
n=1 n=1

µ
Using the orthogonality of the sin(nφ) and µr = µ0 , the resultant set of equations for the coefficients of the
Φi is

    Id −2 
an−1 −µr an−1 µr a−n−1 0 An − 2π a δn,1
 an−1 −an−1 −a−n−1 0   Bn   Id a−2 δn,1 
    2π 
 0 µr bn−1 −µb−n−1 b−n−1   Cn  =  0  ∀ n = 1, 2, 3...
0 bn−1 b−n−1 −b−n−1 Dn 0

This system can be solved with Kramer’s rule, Mathematica or similar. For n 6= 1 all coefficients are zero.
For n = 1 one finds

6
 Id 4µr b2
D1 = −
2π b2 (1 + µr )2 − a2 (1 − µr )2

and

Id sin φ 4µr b2
Φ3 = − f with f=
2πρ b2 (1 + µr )2 − a2 (1 − µr )2

Thus, the field is attenuated by the factor f , q.e.d. (No comparison with problem 5.14 required.)

c): The exact field reduction factor for µr = 200, b = 12.5mm and wall thickness t = b − a = 3mm is
f = 4.56%. For µr À 1 and b À t it is

2b
f≈ ,
µr t

which yields f ≈ 4.17%.

7
3. Jackson, Problem 5.19 8 Points

a): Since there is no free currents, we use the magnetostatic potential. The potential of the described object
is found from its volume magnetic charge density ρM = −∇ · M = 0 and its surface magnetic charge density
σM = n̂ · M = ±M0 at z = ±L/2, respectively:

Z Z
1 ρM (x0 ) 3 0 1 σM (x0 ) 3 0
ΦM = d x + d x
4π V \∂V |x − x0 | 4π ∂V |x − x0 |

In the given case, on the z-axis the potentials due to the top (T) and bottom (B) surfaces are

Z a
M0 1
ΦT /B = ± 2πρdρ q¡ ¢2
4π ρ=0 ρ2 + z ∓ L2
s 
µ ¶2 a
M0  ρ2 + z ∓ L 
= ±
2 2
0
s 
µ ¶2 ¯ ¯
M0 L ¯ L ¯
= ±  a2 + z ∓ − ¯¯z ∓ ¯¯
2 2 2

(upper signs for T, lower signs for B). The total potential ΦM = ΦT + ΦB , which is

s  
µ ¶2 s µ ¶2  L , z > L/2
M0  L L + M 0
ΦM (z) = a2 + z − − a2 + z + × 2z , |z| ≤ L/2
2 2 2 2 
−L , z < −L/2

On the z-axis, the only non-zero component of H is

Hz = −∂z ΦM (z)
 
½
M0  z − L2 z + L2 0 , |z| > L/2
= − q − q  − M0 ×
2 ¡ ¢
L 2
¡ ¢
L 2 1 , |z| ≤ L/2
a2 + z − 2 a2 + z + 2

On the z-axis, the only non-zero component of B is

µ ½ ¶
0 , |z| > L/2
Bz = µ0 Hz + M0 ×
1 , |z| ≤ L/2
 
µ0 M0  z − L2 z + L2
= − q − q 
2 ¡ ¢2 ¡ ¢
L 2
a2 + z − L2 a2 + z + 2

8
b):

Figure 2: Hz and Bz vs. z/L for L = 5a.