Electrodynamic spherical harmonic

arXiv:0803.4093v1 [math-ph] 28 Mar 2008

A V Novitsky
Department of Theoretical Physics, Belarusian State University,
Nezavisimosti Avenue 4, 220050 Minsk, Belarus
[email protected]

Abstract
Electrodynamic spherical harmonic is a second rank tensor in three-
dimensional space. It allows to separate the radial and angle variables
in vector solutions of Maxwell’s equations. Using the orthonormaliza-
tion for electrodynamic spherical harmonic, a boundary problem on a
sphere can be easily solved.

1 Introduction
In this paper we introduce new function — electrodynamic spherical har-
monic. It is represented as a second rank tensor in three-dimensional space.
But the function differs from tensor spherical harmonic [1–5]. By its defini-
tion the electrodynamic spherical function possesses a number of properties
of usual scalar and vector harmonics and includes them as component parts.
Why electrodynamic spherical harmonic? The word “electrodynamic” im-
plies that the function is applied for solution of vector field problems. We
use it in electrodynamics, however one can apply the function for description
of spin fields in quantum field theory.
Electrodynamic spherical harmonic is not a simple designation of the well-
known functions. It satisfies the Maxwell equations and describes the angular
dependence of vector fields. The introduced function separates the variables
(radial coordinate and angles) in the fields. Moreover, the notation of the
fields in terms of electrodynamic spherical harmonic noticeably simplifies the
solution of a boundary problem on spherical interface. Just application of

1
the ortonormalization condition allows to find the coefficients of spherical
function expansion (e.g. scattering field amplitudes).

2 Scalar and vector spherical harmonics

Hamilton’s operator ∇ in three-dimensional space contains the derivatives on
three coordinates. In spherical coordinates (r, θ, ϕ) the derivatives on radial
coordinate and angles can be separated. This is achieved by representing the
unit tensor 1 in 3D space as the superposition of two projection operators:
!
r ⊗ r r×2 1 1
∇ = 1∇ = 2
− 2 ∇ = 2 r(r∇) − 2 r× (r× ∇),
r r r r

where r ⊗ r/r 2 is the projector onto the direction er = r/r, −r×2 /r 2 is the
projector onto the plane orthogonal to the unit vector er . Tensor r× dual to
the vector r gives the well-known vector product when acting on a vector a:
r× a = r × a and ar× = a × r [6]. Introducing the vector differential operator
1
L= r×∇ (1)
i
the equation (??) is rewritten as follows
r ∂ i
∇= − 2 r × L. (2)
r ∂r r
Vector L is called orbital angular momentum operator in quantum me-
chanics, because it is presented as vector product of radius vector r and
momentum p = −i∇ operators. L includes only derivatives on the angles θ
and ϕ. Using the definition of L one can derive the Laplace operator
L2
!
1 ∂2 ∂
∆=∇ = 2 r2 − (3)
r ∂r ∂r r2
and the following properties:

rL = 0, (Lr) = 0, L2 L = LL2 , L × L = iL. (4)

Scalar spherical harmonic Ylm (θ, ϕ) is defined as the eigenfunction of the

operator L2 :
L2 Ylm = l(l + 1)Ylm , (5)

2
where l and m are integer numbers. Number m is the eigenvalue of the
operator of projection of angular momentum onto the axis z, Lz :

Lz Ylm = mYlm . (6)

Spherical harmonics Ylm are orthogonal and normalized by the unit:

Z π Z 2π
Yl∗′ m′ (θ, ϕ)Ylm (θ, ϕ) sin θdθdϕ = δl′ l δm′ m . (7)
0 0

where the sign ∗ denotes the complex conjugate.

If we multiply equation (5) by the vector operator L and take into account
the commutation of L and L2 , we will obtain that vector LYlm satisfies the
same equation (5), too. The quantity defined as
1
Xlm = q LYlm (8)
l(l + 1)

is called vector spherical harmonic. The coefficient before LYlm is chosen so

that the orthonormalization condition is of the form
Z π Z 2π
X∗l′ m′ (θ, ϕ)Xlm (θ, ϕ) sin θdθdϕ = δl′ l δm′ m . (9)
0 0

From the self-conjugacy of the angular momentum operator L and prop-

erties (4) the orthogonality
π 2π π 2π Yl∗′ m′ er (L × L)Ylm
Z Z Z Z
er (X∗l′ m′ × Xlm ) sin θdθdϕ = q sin θdθdϕ
0 0 0 0 l(l + 1)l′ (l′ + 1)
π 2π iYl∗′ m′ er LYlm
Z Z
= q sin θdθdϕ =
(10)
0.
0 0 l(l + 1)l′ (l′ + 1)

follows. Below we give some properties of vector spherical harmonics:

q
LXlm = l(l + 1)Ylm , L(er × Xlm ) = 0. (11)

Scalar (vector) spherical harmonics satisfy the scalar (vector) equation for
eigenfunctions of the squared orbital angular momentum operator L2 . The
orthogonality condition for the scalar (7) and vector (9) spherical harmonics
have the same form. Therefore, one can hope to combine them into one
mathematical object.

3
3 Electrodynamic spherical harmonic: defi-
nition and properties
We define an electrodynamic spherical harmonic as a second rank tensor in
three-dimensional space

Flm = Ylm er ⊗ er + Xlm ⊗ eθ + (er × Xlm ) ⊗ eϕ . (12)

The first term of (12) determines the longitudinal part of the tensor. It is
calculated by means of the scalar spherical function. The last two summands
of (12) fix the transverse solution, in the plane (θ, ϕ) perpendicular to the
direction er . It includes vector spherical harmonics. The left and right
vectors in dyads form two sets of orthogonal vectors: (Ylm er , Xlm , er × Xlm )
and (er , eθ , eϕ ).

3.1 Orthonormalization
Multiplying the electrodynamic spherical harmonic by the Hermitian conju-
+
gate Flm we get to

Fl+′ m′ Flm = Yl∗′ m′ Ylm er ⊗er +X∗l′ m′ Xlm (eθ ⊗eθ +eϕ ⊗eϕ )+(er (X∗l′ m′ ×Xlm ))e×
r .
(13)
The quantity before each dyad is orthogonal or normalized as (7), (9), or
(10). Therefore, the orthonormalization condition for electrodynamic spher-
ical harmonics is
Z π Z 2π
+
Flm (θ, ϕ)Flm (θ, ϕ) sin θdθdϕ = 1δl′ l δm′ m . (14)
0 0

Here we consider the vectors er , eθ , and eϕ to be constant in dyads and dual

tensor e×
r . If tensors Flm are the solutions of equations, we can always write
+
these solutions in components. For each component of the tensor Flm Flm the
orthonormalization is carried out. We will obtain the same, if the dependence
of the orts er , eθ , eϕ on angles in (13) is omitted, i.e. they are regarded as
constants. However, when substituting Flm in equations, we should take into
account the angular dependence of the orts.

4
3.2 Explicit form
Let us substitute the explicit expression for the vector operator L
∂ eθ ∂
L = −ieϕ +i (15)
∂θ sin θ ∂ϕ
into equation (12). Then the electrodynamic spherical harmonic is equal to
 !
i ∂ ∂ 
Flm = er ⊗ er + q I − e×
r sin θ Ylm , (16)
sin θ l(l + 1) ∂ϕ ∂θ

where I = −e× ×
r er = 1 − er ⊗ er is the projection operator onto the plane (θ,
ϕ). To calculate the derivatives one can replace them by means of operators
Lz and L± = Lx ± Ly as
∂ ∂ 1
= iLz , = (e−iϕ L+ − eiϕ L− ),
∂ϕ ∂θ 2
because their action on the scalar spherical harmonic is well-known:
q q
L+ Ylm = (l − m)(l + m + 1)Yl,m+1 , L− Ylm =
(l + m)(l − m + 1)Yl,m−1 .
(17)
Hence, the electrodynamic spherical harmonic can be presented as follows
 !
1 i sin θ −iϕ
Flm = er ⊗ er − q ILz + e×r (e L+ − eiϕ L− )  Ylm .
sin θ l(l + 1) 2
(18)
It is easy to exclude the operators from (18). The final formula for tensor
Flm contains scalar spherical harmonics as angle dependence. Unit vectors
er , eθ , eϕ determine the structure of the tensor in three-dimensional space.
Flm is formed by three basic tensors: er ⊗ er , e× r , and I. Hence, it commutes
with each of these tensors.

3.3 Invariants
The first invariant of the electrodynamic spherical harmonic as tensor quan-
tity is its trace. The trace of the tensor (12) equals

tr(Flm ) = Ylm + 2(eθ Xlm ). (19)

5
 The second invariant is determinant

det(Flm ) = Ylm X2lm . (20)

The third invariant of three-dimensional tensor Flm is the trace of the ad-
joint tensor F lm . Adjoint tensor is defined by F lm Flm = Flm F lm = det(Flm )1
and equals

F lm = X2lm er ⊗ er + Ylm eθ ⊗ Xlm + Ylm eϕ ⊗ er × Xlm . (21)

Further the trace is easily calculated:

tr(F lm ) = X2lm + 2Ylm (eθ Xlm ). (22)

Using these three invariants one can find other ones. For example, the trace of
2
squared tensor is determined from equation tr(Flm ) = (tr(Flm ))2 − 2tr(F lm ).

3.4 Generalization of electrodynamic spherical harmonic

The main condition on electrodynamic spherical harmonic is the orthonor-
malization (14). There is more general form of the second rank tensor spher-
ical harmonic satisfying this equation. It is

Glm = Ylm er ⊗ a + Xlm ⊗ b + (er × Xlm ) ⊗ c. (23)

Unit vectors a, b, and c form the orthogonal basis in three-dimensional space.

In equation (12) we have assumed a = er , b = eθ , c = eϕ because of the
spherical symmetry of the function. If we will take a = er , b = eθ , c = −eϕ ,
then the electrodynamic spherical harmonic becomes more complex function
in explicit form, however its invariants are simplified (e.g., the trace equals
tr(Flm ) = Ylm ).

4 Solution of Maxwell’s equations

In this section we will find the spherically symmetric solutions (i.e. electric
E and magnetic H fields) of the Maxwell equations

∇ × E(r) = ikµH(r), ∇ × H(r) = −ikεE(r) (24)

6
for the monochromatic electromagnetic waves in isotropic medium with di-
electric permittivity ε and magnetic permeability µ. The quantity k = ω/c is
called wavenumber in vacuum, and ω is the wave frequency. Arbitrary time
dependence can be obtained by using the linear superposition of monochro-
matic waves E(r) exp(−iωt).
We will search the solution in the form

E(r) = Flm (θ, ϕ)El (r). (25)

The components of the vector El in spherical coordinates depend only on the

radial coordinate r. The dependence on the angle coordinates presents only
in the basis vectors. So, the vector El looks like

El (r) = Erl (r)er + Eθl (r)eθ + Eϕl (r)eϕ . (26)

Further we should calculate rotE. By substituting Hamilton’s operator

(2) one obtains
∂E i ↓ i
∇ × E = e×
r − L(er E) + er (LE), (27)
∂r r r
where the arrow ↓ implies that operator L acts only on vector E, but not er .
Let us calculate each summand of equation (27) applying the solution (25).
The first term is of the form
l
∂E ∂E
e×
r = Flm e×
r , (28)
∂r ∂r
where the commutation relation [Flm , e×
r ] = 0 is taken into account. The
second summand yields
↓ ↓ ↓
L(er E) = L(er E) − L(er E) = L(Ylm Erl ) − L(er E). (29)
↓
The quantity L(er E) is easily calculated using the explicit expression (15) of
the operator L and the relationships ∂er /∂θ = eθ and ∂er /∂ϕ = eϕ sin θ:
↓
L(er E) = −ie×
r E. (30)

So, we get the formula

↓ q 
L(er E) = Flm l(l + 1)eθ ⊗ er + ie×
r El . (31)

7
The third term in (27) can be rewritten using the equation (11):
  q
er (LE) = er Erl L(Ylm er ) + Eθl LXlm + Eϕl L(er × Xlm ) = l(l + 1)Flm (er ⊗eθ )El .
(32)
Finally, the curl of the electric field vector E equals
 q 
∂ 1 i l(l + 1)
∇ × E = Flm (θ, ϕ) e× + e× − e×  El (r). (33)
r
∂r r r r ϕ

In expression (33) we have took into account the derivatives on the angle
variables. Therefore, further the orts of spherical coordinates er , eθ , and eϕ
can be considered as constants. The Maxwell equations are reduced to the
set of ordinary differential equations
q
dEl 1 × l i l(l + 1) × l
e×
r + er E − eϕ E = ikµHl ,
dr r q r
dH l
1 l
i l(l + 1) × l
e×
r + e× r H − eϕ H = −ikεEl . (34)
dr r r
Equations (34) allow to determine the radial dependence of the fields.
Multiplying the set of equations (34) by the unit vector er we can express
the longitudinal components of the fields as follows
q q
l(l + 1) l(l + 1)
Erl = − Hθl , Hrl = Eθl . (35)
εkr µkr
The tangential field components Elt = IEl = Eθl eθ + Eϕl eϕ and Hlt = IHl are
determined from the equations which can be written in matrix form:
d(rW)
= ikM(rW), (36)
dr
where
! !
Hlt 0 εA l(l + 1)
W= , M= , A = e×
r − eϕ ⊗ eθ .(37)
Elt −µA 0 εµk 2 r 2

Equation (36) is satisfied for inhomogeneous media ε(r), µ(r), too. Such
matrix equation can be solved numerically for arbitrary medium, or analyti-
cally for homogeneous one. Let us find tangential field components W when

8
ε = const and µ = const. The simplest way is to write the equation for
projection Wθ = (Hθl , Eθl ):
d2 (rWθ )
!
l(l + 1)
2
+ k 2 εµ − (rWθ ) = 0, (38)
dr r2
The solutions of the equation (38) are well-known and can be presented as
! ! !
(1) c1 (2) c2 (1) (2) eθ
Wθ = f +f = (f c1 + f c2 ) , (39)
c′1 c′2 eϕ
where c1 and c2 are constant vectors. The couples of independent solu-
tions are spherical Bessel functions f (1) = jl (nkr), f (2) = j−l−1 (nkr) or
(1)
spherical Hankel functions of the first and second kind f (1) = hl (nkr),
(2) √
f (2) = hl (nkr). n = εµ is the refractive index. After determining the
ϕ-projections of the fields as
!
i 0 −ε d(rWθ )
Wϕ = (40)
krεµ µ 0 dr
one can write the transverse vector field
! !
η1 (r) η2 (r) c1
W(r) = , (41)
ζ1 (r) ζ2 (r) c2
where
i d(rf (1,2) )
η1,2 = f (1,2) eθ ⊗ eθ − eϕ ⊗ eϕ ,
µkr dr
(1,2) i d(rf (1,2) )
ζ1,2 = f eθ ⊗ eϕ + eϕ ⊗ eθ . (42)
εkr dr
Tangential field vector W plays an important part, because it is contin-
uous on the spherical surface. That is why it can be applied for the study of
electromagnetic wave diffraction by a sphere.

5 Conclusion
Thus, the general solution of the Maxwell equations is of the form
l
∞ X
! ! !
H(r) X
l η1l (r) η2l (r) cl1
= Flm (θ, ϕ)V (r) , (43)
E(r) l=0 m=−l
ζ1l (r) ζ2l (r) cl2

9
where V l is the matrix that takes into account the longitudinal components
of electric and magnetic fields. This matrix can be easily calculated from
the equation (35). In each partial solution included into the general one (43)
the radial and angle variables are separated. Using the orthonormalization
(14) for the electrodynamic spherical harmonic Flm , each partial wave can
be easily singled out:
! ! !
η1l (r) η2l (r) cl1 π 2π H(r, θ′ , ϕ′ )
Z Z
l +
V (r) = Flm (θ′ , ϕ′ ) sin θ′ dθ′ dϕ′ .
ζ1l (r) ζ2l (r) cl2 0 0 E(r, θ′ , ϕ′ )
(44)
This property of the electrodynamic spherical harmonic is very useful for the
investigation of electromagnetic wave scattering. In scattering, the boundary
condition is the single equation for tangential fields W. It is easily solved,
if the orthonormalization is applied. Some attempts of investigation of scat-
tering in the similar manner as described above have been made in [7].
In the general solution (43) the constants c1,2 determined by initial condi-
tions are explicitly shown. Vectors c1 and c2 set independent solutions. For
instance, if c1 = 0, then the radial dependence is determined by the function
f (2) (r), and vice versa. If c2 = 0 and f (1) (r) = h(1) (nkr), then the field (43)
is the multipole expansion [8]. The amplitude of electric multipole field is
equal to aE (l, m) = c1 eθ , the amplitude of magnetic multipole field is equal
¯
to aM (l, m) = c1 eϕ . So, vector c1 can be called vector amplitude of multipole
fields.
In further investigations we will study the scattering and multipole ex-
pansion of electromagnetic fields in details.

References
References
[1] Newman E and Penrose R 1966 Note on the Bondi-Metzner-Sachs group
J. Math. Phys. 7 836–870

[2] James R W 1976 New Tensor Spherical Harmonics, for Application to

the Partial Differential Equations of Mathematical Physics Phil. Trans.
R. Soc. London. Series A 281 195–221

10
[3] Sandberg V D 1978 Tensor spherical harmonics on S 2 and S 3 as eigen-
value problems J. Math. Phys. 19 2441–2446

[4] Winter J 1982 Tensor spherical harmonics Letters in Mathematical

Physics 6 91–96

[5] Kostelec P, Maslen D K, Rockmore D N, and Healy Jr D 2000 Computa-

tional harmonic analysis for tensor fields on the two-sphere J. Comput.
Phys. 162 514–535

[6] Fedorov F I 1976 Theory of Gyrotropy (Minsk: Nauka i Tehnika)

[7] Novitsky A V and Barkovsky L M 2006 Evolution and impedance op-

erators of spherically symmetric bianisotropic media J. Phys. A: Math.
Gen. 39 7543-60

[8] Jackson J D 1998 Classical electrodynamics (New York: John Wiley &
Sons)

11