Using the integral representation for the translation coe cients, making the transformation ϕ0 → ϕ0 + π, changing the variable of integration from β to π − β and using the identities (A.24) and (A.25), yields
Amn,m n (−kr0) = (−1)n+n Amn,m n (kr0) ,
Bmn,m n (−kr0) = (−1)n+n +1Bmn,m n (kr0) .
Further, since
Amn,m n (kr0) = (−1)n+n A−m n ,−mn (kr0) ,
Bmn,m n (kr0) = (−1)n+n +1B−m n ,−mn (kr0) ,
we obtain
Amn,m n (−kr0) = A−m n ,−mn (kr0
Bmn,m n (−kr0) = B−m n ,−mn (kr0
Recurrence relations for the scalar and vector addition theorem has also been given by Chew [32, 33], Chew and Wang [35] and Kim [117]. The relationship between the coe cients of the vector addition theorem and those of the scalar addition theorem has been discussed by Bruning and Lo [29], and Chew [32].
C
Computational Aspects
in E ective Medium Theory
In this appendix we compute the basic integrals appearing in the analysis of electromagnetic scattering from a half-space of randomly distributed particles. Our derivation follows the procedures described by Varadan et al. [236], Tsang and Kong [223, 226], and Tsang et al. [228].
C.1 Computation of the Integral Imm1 n
The integral Imm1 n is
Imm1 n = ejKe·r0p u3m −mn (ksrlp) dV (r0p) ,
Dp
where Ke = Ksez , and the integration domain Dp is the half-space z0p ≥ 0, excluding a spherical volume of radius 2R centered at r0l. The volume integral can be transformed into a surface integral by making use of the following result. Let u and v be two scalar fields satisfying the Helmholtz equation in the bounded domain D, with the wave numbers Ks and ks, i.e.,
∆u + Ks2u = 0, ∆v + ks2v = 0.
Then, from Green’s theorem we have
|
1 |
|
|
|
|
|
|
|
D uvdV = |
|
D (v∆u − u∆v) dV |
|
ks2 − Ks2 |
|
1 |
|
|
∂u |
|
∂v |
|
|
= |
|
|
v |
|
− u |
|
dS, |
|
ks2 − Ks2 |
S |
∂n |
∂n |
where S is the boundary surface of the domain D, and n is the outward unit normal vector to S. Using this result we transform the integral Imm1 n as follows:
290 C Computational Aspects in E ective Medium Theory
z
Fig. C.1. Integration surfaces SR , S∞ and Sz
1 |
|
1 |
|
|
3 |
|
|
|
|
∂ |
e |
jK |
r |
|
Imm n = |
|
|
|
|
|
|
um −mn (ksrlp) |
|
|
e· |
0p |
k2 |
− |
K2 |
|
∂n |
|
|
s |
s SR S∞ Sz |
|
|
|
|
|
|
|
|
|
|
|
− |
ejKe·r0p |
∂um3 −mn |
(k r |
|
) dS (r |
|
) . |
|
|
|
|
|
|
|
∂n |
|
s |
lp |
|
0p |
|
|
|
|
|
|
The integral can be decomposed into three integrals
Imm1 n = Imm1,R n + Imm1,∞ n + Imm1,z n ,
where Imm1,R n is the integral over the spherical surface SR of radius 2R cen-
1,∞
tered at r0l, Imm n is the integral over the surface of a half-sphere S∞ with radius R∞ in the limit R∞ → ∞, and Imm1,z n is the integral over the xy- plane Sz (the plane z = 0). The choice of the integration surfaces is shown in Fig. C.1.
Using the identity r0p = r0l + rlp and replacing, for convenience, the variables rlp, θlp and ϕlp by r, θ and ϕ, respectively, we obtain
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Imm1,R |
n = − |
1 |
|
ejKsz0l |
|
|
|
|
|
|
|
|
|
|
ks2 − Ks2 |
|
|
|
|
|
|
|
|
|
|
|
|
|
!h(1) |
(ksr) P |m −m| |
(cos θ) ej(m −m)ϕ |
∂ |
ejKe·r |
|
|
|
|
|
|
× SR |
|
n |
|
|
n |
|
|
|
∂r |
|
|
− |
ejKe·r |
∂ |
|
h(1) (ksr) P |m −m| (cos θ) ej(m −m)ϕ" dS(r), |
|
∂r |
|
|
|
|
n |
|
|
|
n |
|
|
|
|
|
|
whence, taking into account the series representation |
|
|
|
|
|
|
∞ |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2n |
+ 1 |
|
|
|
|
|
ejKe·r = |
2jn |
|
|
|
|
|
|
jn (Ksr) Pn (cos θ) |
(C.1) |
|
|
2 |
|
|
|
|
|
|
n =0 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
C.1 Computation of the Integral Imm1 n |
291 |
for Ke · r = Ksr cos θ, the orthogonality of the associated Legendre functions and the relation Ke · r0l = Ksz0l, we end up with
1,R |
16πR3 |
|
2n + 1 |
|
|
|
jn |
|
|
|
|
ejKsz0l δmm Fn (Ks, ks, R) , |
Imm n = |
|
|
|
|
|
|
2 |
2 |
|
2 |
|
where |
(ksR) − (KsR) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
(1) |
|
Fn (Ks, ks, R) = (ksR) hn |
(2ksR) jn (2KsR) |
− (KsR) h(1)n (2ksR) (jn (2KsR)) .
1,∞
To compute Imm n we use the stationary point method. Using the inequality Im{Ks} > 0 and the asymptotic expressions of the spherical Hankel functions of the first kind
(−j)n +1ejksr
ksr
,
= j(−j)n +1ejksr , r
1,∞
we see that Imm n vanishes as R∞ → ∞.
To evaluate Imm1,z n we pass to cylindrical coordinates, integrate over ϕ, and obtain
1,z |
|
|
|
|
|
|
2π |
|
∞ |
(1) |
|
|
|
|
|
|
|
|
Imm n |
= − |
|
|
|
|
|
|
|
δmm |
|
jKshn |
(ksr) Pn (cos θ) |
k2 |
− |
K2 |
|
|
|
|
|
|
|
s |
|
s |
|
0 |
|
|
|
|
|
|
" |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
∂ |
|
(1) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
− |
|
|
|
hn (ksr) Pn (cos θ) |
|
|
|
|
ρ dρ. |
|
|
|
∂z |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
z=−z0l |
|
Further, using Kasterin’s representation [228, 258] |
|
|
|
|
|
|
|
|
(1) |
|
|
|
|
|
|
|
|
|
|
|
|
|
n |
|
1 |
|
∂ |
(1) |
|
|
|
hn (ksr) Pn (cos θ) = (−j) |
Pn |
|
|
|
|
|
|
h0 (ksr) |
jks |
∂z |
with |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
(1) |
(ksr) = |
|
ejksr |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
h0 |
|
|
|
|
|
, |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
jksr |
|
|
|
|
|
|
|
|
|
|
we find that |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1,z |
2π |
|
|
|
|
|
|
n |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Imm n = − |
|
|
|
|
(−j) δmm |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
k2 |
− |
K |
2 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
s |
|
s |
|
|
|
|
|
1 ∂ ejks√ |
|
|
|
|
|
∞ |
|
|
|
|
|
|
|
|
|
|
|
|
∂ |
ρ2+z2 |
× 0 |
|
|
|
jKs − |
|
|
Pn |
|
|
|
|
|
|
|
|
|
ρ dρ. |
|
|
|
∂z |
|
jks |
∂z |
|
jks |
|
|
|
|
|
|
|
|
|
ρ2 + z2 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
z=−z0l |