PHYSICAL REVIEW B 107, 155425 (2023)

Electromagnetic scattering by arbitrary-shaped magnetic particles and multipole decomposition:

Analytical and numerical approaches
1,2,*
Andrey B. Evlyukhin and Vladimir R. Tuz3,4
1
Institute of Quantum Optics, Leibniz University Hannover, Welfengarten 1, 30167 Hannover, Germany
2
Cluster of Excellence PhoenixD, Leibniz University Hannover, Welfengarten 1A, 30167 Hannover, Germany
3
State Key Laboratory on Integrated Optoelectronics, College of Electronic Science and Engineering, International Center of Future Science,
Jilin University, 2699 Qianjin Street, Changchun 130012, China
4
School of Radiophysics, Biomedical Electronics and Computer Systems, V. N. Karazin Kharkiv National University, 4, Svobody Square,
Kharkiv 61022, Ukraine

(Received 28 December 2022; revised 12 March 2023; accepted 5 April 2023; published 21 April 2023)

We develop a theoretical approach that makes it possible to analyze the role of multipole contributions in
the scattering spectra of arbitrarily shaped magnetic particles with the relative permittivity and permeability
given in a general tensor-valued form. The method for calculating the exact multipole moments with the
inclusion of induced electric polarization and magnetization is suggested and explicitly applied for dipole and
quadrupole terms. The final expressions for the dipole and quadrupole moments are obtained in such a form
that they can be easily implemented in numerical solvers. Successful verification of the analytical expressions
for multipole moments and the multipole decomposition of the scattered field and scattering cross section is
provided by comparing the results obtained for differently shaped magnetic particles using the analytical Mie
theory, numerical discrete dipole approximation, and COMSOL MULTIPHYSICS software. As a particular example,
the manifestation of the Faraday effect for fields scattered by a ferrite sphere is discussed in the framework
of the multipole decomposition method with the derived magnetization terms. Accounting for the magnetic
properties of particles in the multipole analysis of the electromagnetic scattering significantly expands the
method’s capabilities for studying and modeling metamaterial properties in the spectral ranges where natural
materials can have a relative permeability different from unity.

DOI: 10.1103/PhysRevB.107.155425

I. INTRODUCTION metamaterial band-stop filters [7–9]. Magnetic materials are

usually integrated into traditional passive metamaterials as
All-dielectric metamaterials are a subclass of artificial
a substrate or superstrate to construct polarization selective
media that are composed of high refractive index particles
devices and perfect microwave absorbers [10,11]. Fabrica-
(resonators) without any metallic inclusions [1]. The physics
tion of resonators directly from magnetic materials opens up
of dielectric metamaterials is related to wave scattering mech-
new possibilities for the development of all-ferrodielectric
anisms, Mie resonances, and interference effects that arise
metamaterials [12,13]. This greatly relies on a theoretical
between various multipoles generated in the constitutive par-
understanding of effects, as well as using electromagnetic
ticles by incoming radiation. Further progress in developing
theory and modeling approaches to enable quantitatively re-
advanced tunable metamaterials partly depends on introduc-
liable predictions for ever more complex metamaterials and
ing new substances into their compositions.
device geometries [14,15]. In addition, the exotic effects
Given the development of metadevices [2], introducing
caused by various resonant states discovered in all-dielectric
magnetic materials into metamaterial compositions allows
metamaterials, such as toroidal [16–18], anapole [19,20], and
one to enhance magneto-optical effects and develop active
supercavity [21] modes, still require detailed theoretical and
metamaterials in which magnetic field control provides an
experimental investigations in configurations based on mag-
effective way of achieving tunable designs [3]. Magnetic
netic materials.
metamaterials demonstrate promising potential for efficient
It is important to note that many new materials used in
and nonreciprocal control of wave propagation by magne-
metamaterial compositions exhibit natural anisotropy. More-
tization [4–6]. Typical magnetic materials are ferrites and
over, when applying external influences to metamaterials,
their derivatives, which exhibit ferromagnetic resonance un-
such as the static magnetic field, gyrotropy and related
der external bias magnetic field. Ferrites are widely used for
effects may occur. This requires the corresponding devel-
creating tunable devices in the microwave region, including
opment of methods for studying metamaterials containing
negative refraction metamaterials, metamaterial antennas, and
particles made of anisotropic substances. Specifically, to
study scattering by anisotropic particles, many numerical
approaches have been proposed so far. They are the Mie
*
[email protected] theory [22–25], method of moments [26], discrete dipole

2469-9950/2023/107(15)/155425(13) 155425-1 ©2023 American Physical Society

ANDREY B. EVLYUKHIN AND VLADIMIR R. TUZ PHYSICAL REVIEW B 107, 155425 (2023)

approximation (DDA) [27,28], finite-difference time-domain

method [29,30], T -matrix method [31,32], integral equa-
tions [33,34], and others (see a review on methods in
Ref. [35]).
To design metamaterials with prescribed electromagnetic
properties, one should select an elementary unit cell of the
structure (often called the meta-atom) and optimize its scat-
tering characteristics by adjusting geometric and material
parameters. For an individual particle, this can be done by
numerically solving the Maxwell equations for the fields and
applying the multipole expansion to determine the multipoles
contributing to the scattering [36]. In optics, multipole expan- FIG. 1. Schematic illustration of a scattering system: an incident
sion is widely used to describe light scattering by dielectric electromagnetic wave with wave vector k and electric field E0 , prop-
particles, whereas it is quite rarely used for magnetic ones. agating in a medium with constant and scalar permittivity εs and
This is partly due to the fact that in the optical range, the permeability μs , is scattered by a finite-sized object with volume Vp
materials are nonmagnetic substances with relative permeabil- and tensor-valued permittivity ε̂ p and permeability μ̂ p .
ity equal to unity, whereas multipole expansions are related
to behaviors of only electric charges producing the electric
currents. Contrariwise, the specific properties of magnetic ma- 
ε̂ p − 1̂εs , r ∈ Vp ,
terials are manifested mainly in the microwave range, and the ε̂ = (4)
magnetic charges (monopoles) formally cannot be considered 0, r∈/ Vp ,
since they do not physically exist. Nevertheless, if we assume Eqs. (1) and (2) can be rewritten as
that each particle possesses fictitious magnetic charges (and
thus fictitious magnetic currents) that can be introduced as ∇ × E(r) = iωμ0 (1̂μs + μ̂) H(r), (5)
a divergence of its magnetization, then the formalism of the
multipole expansion can also be applied to describe the prop- ∇ × H(r) = −iωε0 (1̂εs + ε̂) E(r), (6)
erties of magnetic particles [37].
In the present paper, our goal is to derive expressions for where Vp is the volume of the scatterer, 1̂ is the 3 × 3 unit
the exact multipole moments in the case of wave scattering tensor, and i is the imaginary unit.
by magnetic particles of arbitrary shape and size. The final Wave equations are obtained from Eqs. (5) and (6) by
expressions for the dipole and quadrupole moments are ob- applying the operation ∇× to them and excluding H and
tained in such a way that they can be easily implemented in E from the first and second equations, respectively. These
numerical solvers. To confirm this, we implement the obtained manipulations yield:
expressions for the multipole moments into the calculation ∇ × ∇ × E − k 2 E = ω2 μ0 μs P + iωμ0 ∇ × M, (7)
scheme of the DDA method [38–41] and verify the results by
comparing the scattering characteristics of a spherical mag- ∇ × ∇ × H − k 2 H = ω2 ε0 εs μ0 M − iω∇ × P, (8)
netic particle with those obtained with the Mie method. Then, √
for spherical and disk-shaped particles, a manifestation of the where k = k0 εs μs and k0 is the vacuum wave number. Here
Kerker (backscattering suppression) [42] and Faraday effects we introduced the vectors of polarization and magnetization,
is discussed. Moreover, for magnetic particles with a complex
shape, the results are checked against full-wave simulations P(r) = ε0 ε̂ E(r), M(r) = μ̂ H(r), (9)
using the commercial COMSOL MULTIPHYSICS solver. respectively.
Using the Green’s tensor Ĝ(r, r ) for a homogeneous
II. GENERAL ANALYTICAL APPROACH medium with εs and μs , which satisfies the tensorial equation
The first two Maxwell equations for monochromatic elec- ∇ × ∇ × Ĝ(r, r ) − k 2 Ĝ(r, r ) = 1̂δ(r − r ), (10)
tric E and magnetic H fields in a system with tensor-valued
relative permittivity ε̂(r) and permeability μ̂(r), depending on one can write the scattered electric field Escat as
the position in the system, are written as 
Escat (r) = ω μ0 μs
2
ĜFF (r, r )P(r )dr
∇ × E(r)e−iωt = iωμ0 μ̂(r)H(r)e−iωt , (1) Vp

∇ × H(r)e−iωt = −iωε0 ε̂(r)E(r)e−iωt , (2) + iωμ0 ĜFF (r, r )[∇ × M(r )]dr (11)
Vp
where r is the radius vector of a point in the system, ω is the and the scattered magnetic field Hscat as
angular frequency, t is the time, and ε0 and μ0 are the vacuum 
permittivity and permeability, respectively. For the scattering
Hscat (r) = ω2 ε0 εs μ0 ĜFF (r, r )M(r )dr
system shown in Fig. 1, by introducing the tensors Vp
 
μ̂ p − 1̂μs , r ∈ Vp ,
μ̂ = (3) − iω ĜFF (r, r )[∇ × P(r )]dr , (12)
0, r∈/ Vp ,
Vp

155425-2
ELECTROMAGNETIC SCATTERING BY ARBITRARY- … PHYSICAL REVIEW B 107, 155425 (2023)

where GFF (r, r ) is the far-field approximation of Ĝ(r, r ) and the extinction power Wext can be found from the optical
[43], which is determined by theorem [39]
 
eikr ω 1 ik
GFF (r, r ) =

(1̂ − nn)e−ik(n·r ) , (13) Wext = Im E0∗ · p + [m × ninc ] − (Q̂ninc )
4π r 2 v 2

where nn is the dyadic product of the unit vector n = r/r ik
+ [ninc × (M̂ninc )] + · · · , (20)
directed to the observation point from the origin of the co- 2v
ordinate system.
where E0 is the electric field at the point of the multipole
moment localization corresponding to the incident electro-
III. MULTIPOLE DECOMPOSITION magnetic plane wave Eeik(ninc ·r) and the asterisk denotes the
To implement the multipole decomposition for the scat- complex conjugation.
tered field, we introduce the total induced current density
A. Multipole moments
1
jt (r) = −iωP(r) + ∇ × M(r) (14) Expressions for calculating multipole moments related
μs
to the scattered fields are determined by the method of
and rewrite Eq. (11) in an ordinary representation [40]: their introduction. There are two general approaches: (i) the
 long-wavelength approximation (LWA) based on the Taylor
Escat (r) = iωμ0 μs ĜFF (r, r ) jt (r )dr . (15) expansion of the electromagnetic potentials [41] and (ii) the
Vp exact multipole approach (EMA) based on the spherical har-
From this representation, we can derive the multipole decom- monics presentation of the scattered fields [40,44]. The LWA
position of the scattered field in an ordinary form [40]: is better suited for the cases where the scattering object is
 smaller than the incident wavelength and the main multipole
k 2 eikr 1 moments, determining that the scattered waves can be rep-
Escat (r) = [n × [p × n]] + [m × n]
ε0 εs 4π r v resented by only their first quasistatic approximation. As the
 size of the scatterer increases, the number of such quasistatic
ik ik
+ [n × [n × (Q̂n)]] + [n × (M̂n)] + · · · , multipole moments (including dipole terms) increases rapidly,
2 2v making multipole analysis unnecessarily complex. In the case
(16) of EMA, the multipole moments are described by exact in-
tegral expressions that give their exact values for scatterers
where n = r/r is again the unit vector directed to the obser- of arbitrary shape and size. Importantly, both mentioned ap-
vation point from the multipole localization point coinciding
√ proaches are mutually related: LWA can be obtained from
with the origin of the coordinate system; v = 1/ ε0 μ0 εs μs is EMA by performing simple Taylor expansions of the spherical
the light velocity in the surrounding medium; and the vector Bessel functions which are included in the definition of exact
of an electric dipole (ED) moment p, the vector of a magnetic multipole moments (see Ref. [44]).
dipole (MD) moment m, the tensor of an electric quadrupole The expressions for the first important multipole moments
(EQ) moment Q̂, the tensor of a magnetic quadrupole (MQ) are presented in Table I. They are related to the polarization
moment M̂, and all other multipoles are determined by the P and magnetization M expressed by the current density
total induced current density given by Eq. (14). Since the total j [40,44] by replacing j with −iωP and [∇ × M]/μs , re-
current density jt includes individual contributions from the spectively. In the latter case of magnetization, subsequent
electric polarization P and magnetization M, all multipole appropriate transformations are applied to eliminate the nabla
moments can also be divided into the electric and magnetic operator ∇ from the integral expressions (see Appendix A).
parts, respectively: The expressions for the multipole moments in LWA are pre-
p = pE + pH , m = mE + mH , (17) sented in Table II.
Note that corresponding expressions for the higher-order
Q̂ = Q̂E + Q̂H , M̂ = M E + M̂ H , (18) multipole moments can be obtained with the same method
using their representation in terms of the current density [40].
where the superscripts E and H are related to the multipole
moments of the polarization P and the magnetization M, B. Point of multipole localization
respectively.
The total scattered power W is then determined by a stan- Note that the above equations, including those collected
dard procedure [40,43], in Tables I and II, refer to multipole moments located at the
origin of the coordinate system. However, it is known that the
k4v k6v  value of multipole moments depends on the choice of their
W = |p|2 + |Qβγ |2
12π ε0 εs 160π ε0 εs μs βγ localization point [45]. Therefore, the question of where to set
the origin for the coordinate system arises. In this regard, there
k4 k6  is an important rule: the point of localization of multipole
+ |m|2 + |Mβγ |2 + · · · , moments must be chosen so that the number of multipoles
12π ε0 εs v 160π ε0 εs μs v βγ
required for the correct description of the scattered fields is
(19) minimal. For homogeneous scatterers of an arbitrary shape,

155425-3
ANDREY B. EVLYUKHIN AND VLADIMIR R. TUZ PHYSICAL REVIEW B 107, 155425 (2023)

TABLE I. Exact multipole moments, including electric dipole (ED), magnetic dipole (MD), electric quadrupole (EQ), and magnetic
quadrupole (MQ) terms. All expressions include the corresponding spherical Bessel function jl (kr) of the lth order.

Moment In terms of polarization P = ε0 (ε̂ p − εs 1̂)E In terms of magnetization M = (μ̂ p − μs 1̂)H

 2  ik 2
 5ik 2
 j2 (kr)
ED pE = Vp j0 (kr)Pdr + k2 Vp j(kr) 2 [3(r · P)r − r P]dr
2 (kr) 2
pH = ωμ s Vp
j1 (kr)
kr
[r × M]dr + 2ωμ s Vp (kr)2
[r × M]dr
4 
− 2ωμ ik
s Vp (kr)3
r [r × M]dr
j3 (kr) 2

  3k 2
 j2 (kr)
MD mE = −iω 23 Vp
j1 (kr)
kr
[r × P]dr m = μs Vp kr Mdr + 2μ
H 3 j1 (kr)
s Vp (kr)2
[r(r · M) − r 2 M]dr
  ik 2
 j2 (kr)
EQ Q̂E = Vp
j1 (kr)
kr
[3(rP + Pr) − 2(r · P)1̂]dr + 2k 2 j3 (kr)
Vp (kr)3
Q̂H = ωμ s Vp (kr)2
{3(r[r × M] + [r × M]r) − 2(r · [r × M])1̂}dr
2ik 2
 j3 (kr)
× [5(r · P)rr − r 2 (Pr + rP) − r 2 (r · P)1̂]dr + ωμs Vp (kr)3 {7(r[r × M] + [r × M]r) + (r · [r × M])1̂}dr
4 
+ 2ik
ωμs Vp (kr)4
j4 (kr)
{5(r · [r × M])rr − r 2 (r[r × M] + [r × M]r)
−(r · [r × M])r 2 1̂}dr
  2 (kr)
MQ M̂ E = −5iω j2 (kr)
Vp (kr)2
([r × P]r + r[r × P])dr M̂ H = μ15s Vp j(kr) 2 [rM + Mr − 3 (r · M)1̂]dr
2

2 
+ 5kμs Vp j(kr) 3 ([r × [r × M]]r + r[r × [r × M]])dr
3 (kr)

this point coincides with their center of mass. Generally, this where δ(r − rl ) is the Dirac δ function, rl is the radius vector
point can be found using various optimization procedures. of the center of the volume discrete cell with the index l, and
pl = ε0 (ε̂ p − εs 1̂)El Vl (22)
IV. CALCULATION OF MULTIPOLE MOMENTS
and
The volume integral expressions collected in Tables I and
II are the definitions of multipole moments. For their calcu- ml = (μ̂ p − μs 1̂)Hl Vl (23)
lation, it is necessary, first, to know the distributions of the are the ED and MD moments corresponding to this cell and
total electric E(r) and magnetic H(r) fields in the scatterers located at rl , with El = E(rl ) and Hl = H(rl ).
and, second, to perform the corresponding integration. In the Typically, in numerical calculations, the choice of the
general case, when the scattering object has a complex shape discretization step (the volume of the discretization cell) is de-
and/or composition, the calculation of full fields can be im- termined by the required accuracy for the result to be obtained.
plemented only numerically using various special software It is important that the total electric and magnetic fields within
packages, including commercially available ones (e.g., COM- the discretization cells can be considered almost uniform.
SOL MULTIPHYSICS). Using representation (21), the integrals from Tables I and
If the total E(r) and H(r) fields inside the scatterer are II can be converted to a simple summation, which can be
known, the integration of the expressions from Tables I and easily implemented numerically. For reference, the discrete
II can be performed by applying a straightforward numerical representations of the exact ED, MD, EQ, and MQ moments
procedure. This implies the discretization of the scatterer on are collected in Table III.
a regular three-dimensional grid with N elementary cells,
where each cell has volume Vl (here l is the index of the V. VERIFICATION PROCEDURES
corresponding cell). This approach allows one to obtain a
discrete representation for the induced polarization P(r) and In this study, we perform the verification of the obtained
magnetization M(r) as follows: representations and demonstrate the capabilities of the devel-
oped approach, considering different examples of scattering

N 
N objects. In particular, we consider differently shaped magnetic
P(r) = pl δ(r − rl ), M(r) = ml δ(r − rl ), (21) particles supporting multipole resonances in the microwave
l=1 l=1 spectral range. Reference [42] showed that for spherical

TABLE II. LWA multipole moments, including electric dipole (ED), magnetic dipole (MD), electric quadrupole (EQ), and magnetic
quadrupole (MQ) terms.

Moment In terms of polarization P = ε0 (ε̂ p − εs 1̂)E In terms of magnetization M = (μ̂ p − μs 1̂)H

 2  
ED pE ≈ Vp Pdr + μ10 sk
Vp
[(r · P)r − 2r 2 P]dr pH ≈ ωμik 2 1
s 2 Vp
[r × M]dr
 k2
 2  2 
MD mE ≈ −iω 21 Vp [r × P]dr + iωμs 20 Vp
r [r × P]dr mH ≈ μ1s Vp Mdr + 10μ k
s Vp
[r(r · M) − 2r 2 M]dr
 2 
EQ Q̂ ≈ Vp [(rP + Pr − 3 (r · P)1̂]dr
E 2
Q̂H ≈ 5ωμ
ik
s Vp
{r[r × M] + [r × M]r − 23 (r · [r × M])1̂}dr
 
MQ M̂ ≈ −iω 31 Vp ([r × P]r + r[r × P])dr
E
M̂ H ≈ μ1s Vp [rM + Mr − 23 (r · M)1̂]dr

155425-4
ELECTROMAGNETIC SCATTERING BY ARBITRARY- … PHYSICAL REVIEW B 107, 155425 (2023)

TABLE III. Discrete representations of the exact multipole moments, including the electric dipole (ED), magnetic dipole (MD), electric
quadrupole (EQ), and magnetic quadrupole (MQ) terms. All expressions include the corresponding spherical Bessel function jl (kr) of the lth
order.

Moment In terms of polarization pl = ε0 (ε̂ p − εs 1̂)El Vl In terms of magnetization ml = (μ̂ p − μs 1̂)Hl Vl

N k2 N ik 2 N 5ik 2 N
ED pE ≈ l=1 j0 (krl )pl + 2
j2 (krl )
l=1 (krl )2 [3(rl · pl )rl − rl2 pl ] pH ≈ ωμs l=1 krl [rl × ml ] + 2ωμs
j1 (krl ) j2 (krl )
l=1 (krl )2 [rl × ml ]
ik 4 N
− 2ωμ s l=1 (krl )3 r [rl × ml ]
j3 (krl ) 2

N N 3k 2 N
MD mE ≈ −iω 23 j1 (krl )
l=1 krl [rl × pl ] mH ≈ 3
μs
j1 (krl )
l=1 krl ml + 2μs
j2 (krl )
l=1 (krl )2 [rl (rl · ml ) − rl2 ml ]
N ik 2 N
EQ Q̂E ≈ l=1 krl [3(rl pl + pl rl ) − 2(rl · pl )1̂] + 2k
j1 (krl ) 2
Q̂H ≈ ωμs l=1 (krl )2 {3(rl [rl
j2 (krl )
× ml ]
N
× l=1 (krl )3 [5(rl · pl )rl rl − rl (pl rl + rl pl ) − rl (rl
j3 (krl ) 2 2
· pl )1̂] + [rl × ml ]rl ) − 2(rl · [rl × ml ])1̂}
2 N
+ 2ik
ωμs l=1 (krl )3 {7(rl [rl × ml ] + [rl × ml ]rl )
j3 (krl )

4 N
+(rl · [rl × ml ])1̂} + 2ikωμs l=1 (krl )4 {5(rl · [rl
j4 (krl )
× ml ])rl rl
−rl2 (rl [rl × ml ] + [rl × ml ]rl ) − (rl · [rl × ml ])rl2 1̂}
N
MQ M̂ E ≈ −iω5 j2 (krl )
l=1 (krl )2 ([rl × pl ]rl + rl [rl × pl ]) M̂ H ≈ μ15s Nl=1 j(kr
2 (krl )
)2
[rl ml + ml rl − 23 (rl · ml )1̂]
l
2 N
+ 5kμs j3 (krl )
l=1 (krl )3 ([rl × [rl × ml ]]rl + rl [rl × [rl × ml ]])

particles with ε p = μ p , the backscattering equals zero re- Ref. [47]). Then, with these moments, the discrete represen-
gardless of the incident wavelength and particle size (the tations of the induced polarization P and magnetization M
Kerker effect). Analytically, in the framework of Mie the- can be obtained from Eqs. (21) (more details on the DDA
ory, this statement is written as an = bn for n = 1, . . . , ∞, numerical method and its applicability for electromagnetic
where an and bn are the scattering coefficients determining, scattering problems by magnetic particles, including the ex-
respectively, the nth-order electric and magnetic multipole pressions for extinction and absorption cross sections, can be
contributions to the scattering cross section σscat
Mie
[46]: found elsewhere [28]).
∞
The total scattering cross section σscat (without applying
2π  the multipole decomposition; labeled “Total” in the figures) is
σscat
Mie
= (2n + 1)(|an |2 + |bn |2 ). (24)
k 2 n=1 defined as [28]
 N
Thus, these contributions of the same-order multipole mo- k4
ments (electric and magnetic dipoles, electric and magnetic σscat = e−ik(n·r j ) {[n × [p j × n]]
16π 2 ε02 εs2 |E0 |2 j=1
quadrupoles, and so on) must coincide with each other in all
spectral ranges where ε p = μ p . 2
We can see the manifestation of the Kerker effect in 1
+ [m j × n]} d , (27)
Fig. 2(a), where the results of the Mie theory calculations for v
the scattering cross section and dipole and quadrupole mo-
ments for a magnetic-dielectric sphere are depicted. In what where E0 is the electric amplitude of the incident wave, n =
follows, we use this effect to validate the correctness of the (sin θ cos ϕ, sin θ sin ϕ, cos θ ), d = sin θ dθ dϕ, and integra-
integral expressions for multipole moments presented above. tion is carried out over the total solid angle . The origin of
Here we consider two numerical verification approaches the Cartesian coordinate system is taken inside the scatterer
based on the use of the DDA (adopted for magnetic particles) in its center of mass. The total scattering cross section as a
and the COMSOL MULTIPHYSICS solver. sum of the multipole terms (labeled “MULT” in the figures) is
obtained from
A. DDA versus Mie theory 2W μ0 μs
σscat = , (28)
The main idea of the DDA method for magnetic particles |E0 |2 ε0 εs
consists of the replacement of the scattering object by a cubic where the scattered power W is given by Eq. (19). Note that
lattice of electric and magnetic point dipoles with known in the DDA, the multipole moments are calculated by direct
electric (α̂ E ) and magnetic (α̂ M ) polarizability tensors: application of the expressions from Table III.
α̂ E = 3ε0 εsVd (ε̂ p − εs 1̂)(ε̂ p + 2εs 1̂)−1 , (25) The results of the DDA calculation for a spherical particle
with μ p = ε p are shown in Fig. 2. We can see good corre-
α̂ M = 3μsVd (μ̂ p − μs 1̂)(μ̂ p + 2μs 1̂)−1 , (26) spondence of the total scattering cross section, the spectral
position of the resonant peaks, and the multipole contributions
where Vd is the volume of the discretization cell. In the follow- with their counterparts obtained from the Mie theory. This
ing consideration, we assume that the scatterer has isotropic confirms the correctness of the expressions derived above
properties with ε̂ p = ε p 1̂ and μ̂ p = μ p 1̂. The corresponding for the multipole terms as well as the applicability of the
electric dipole moment p j and magnetic dipole moment m j suggested numerical method. Importantly, in accordance with
induced in each lattice point j (with the radius vector r j ) are the general theory, full coincidence between contributions
found by solving coupled dipole equations (see Eqs. (1) in of multipole pairs of the same order to the scattering cross

155425-5
ANDREY B. EVLYUKHIN AND VLADIMIR R. TUZ PHYSICAL REVIEW B 107, 155425 (2023)

FIG. 2. Scattering cross sections and corresponding multipole FIG. 3. Scattering cross sections and corresponding multipole
decomposition contributions calculated with the use of (a) Mie the- decomposition contributions calculated using the DDA method for
ory and (b) the DDA method for a magnetic-dielectric sphere with a a magnetic-dielectric disk with a radius of ∼90 mm, a height of
radius of ∼61 mm and constitutive parameters μ p = 5 and ε p = 5. ∼70 mm, and constitutive parameters combinations: (a) μ p = 4 and
The sphere is placed in an airlike environment (εs = μs = 1) and ε p = 8 and (b) μ p = 8 and ε p = 4. The disk is placed in an airlike
irradiated by a linearly polarized plane wave. The inset demon- environment (εs = μs = 1) and irradiated by a linearly polarized
strates the discrete presentation of the sphere applied for the DDA plane wave, as shown in the inset.
calculation.

section is numerically obtained (see the contributions of the either μ p /ε p = 1/2 or μ p /ε p = 2/1 [see Figs. 3(a) and 3(b),
ED-MD and EQ-MQ pairs in Fig. 2). Note that the same respectively]. Since the refractive index for the two combi-
behavior of corresponding multipole pairs is also inherent in nations is fixed, the scattering characteristic (total scattering
particles of another shape with axial symmetry (e.g., disks) if cross section) does not depend on the chosen μ p /ε p ratio.
ε p = μ p [42]. However, the contributions of the corresponding multipoles
to the resonances acquire the rearrangement (ED ⇔ MD,
EQ ⇔ MQ, etc.) on the wavelength scale, as the principle of
B. Multipole duality electromagnetic duality suggests. This result confirms the cor-
Here we perform our calculations for a disk-shaped mag- rectness of the obtained expansions since the duality principle
netic particle under the frontal irradiation conditions of the implies the rearrangement of the spectral hierarchy of reso-
incident wave (i.e., the wave is propagated along the axis of nant contributions of same-order multipoles when the μ p /ε p
the rotational symmetry √ of the disk). We fix the refractive ratios are interchanged.
√
index nR = μ p ε p = 32 and the size of the particle, and We stress that the interchange of values of μ p and ε p leads
two particular combinations of its constitutive parameters are to the rearrangement of the spectral hierarchy of resonant
considered: the ratios between μ p and ε p are chosen to be contributions of same-order multipoles. This follows, for in-

155425-6
ELECTROMAGNETIC SCATTERING BY ARBITRARY- … PHYSICAL REVIEW B 107, 155425 (2023)

Using the total electric E and magnetic H fields in the

scatterer calculated by built-in COMSOL MULTIPHYSICS facil-
ities, the induced polarization P and magnetization M can
be evaluated using Eqs. (9). Then, the multipole moments
can be calculated on the basis of the expressions collected
in Table I. Before using the corresponding expressions, the
time dependence in the form of e+iωt that is accepted in the
COMSOL MULTIPHYSICS environment should be accounted for.
Therefore, to implement the expressions for multipole mo-
ments from Table I in COMSOL MULTIPHYSICS, it is necessary
to change the sign of all imaginary units that are explic-
itly included in them since the other time factor (e−iωt ) has
been taken in our problem statement from the beginning [see
Eqs. (1) and (2)]. The procedure of changing the sign of all
imaginary units is necessary to provide the correct computa-
tion of the simultaneous electric and magnetic contributions
to each multipole moment. Note that the multipole represen-
tations of the total scattered power [Eq. (19)] and scattering
FIG. 4. Separate contributions of the magnetic and electric parts cross section [Eq. (28)] are independent of this peculiarity, in
of total ED = EDE + EDH and MD = MDE + MDH moments in the contrast to the expression for the scattered electric field. For
scattering cross section presented in Fig. 2(b). the latter, the corresponding equation acquires the form

k 2 e−ikr 1
Escat (r) = [n × [p × n]] + [m × n]
stance, from a comparison of the ED and MD contributions to ε0 εs 4π r v

the resonant peaks of scattering functions presented in Fig. 3. ik ik
This behavior is preserved for all multipole pairs of the same − [n × [n × (Q̂n)]] − [n × (M̂n)] + · · · .
2 2v
order.
Thus, the developed approach gives us the ability to study (29)
the multipole compositions of the scattering resonances and
As a basic project for further modification, we adopt
use the obtained information to tune the electromagnetic
the model [48] from the COMSOL MULTIPHYSICS Application
response of magnetic particles. Moreover, in the case of
Gallery. The computational domain of the model includes a
magnetic scatterers, every excited multipole has contributions
dielectric particle placed in a surrounding space (air) truncated
from both electric-induced and magnetic-induced currents.
by a perfectly matched layer (PML). The PML is added to the
Our method allows us to get information about the contri-
model to simulate open (infinite) space and prevent unwanted
butions of these two currents, as shown in Fig. 4 for the
rereflections within the area. Then, Maxwell’s equations, in-
dipole terms of the disk-shaped particle. We can see that
tended for the scattering of electromagnetic radiation with the
the ED (MD) moment of the disk is basically determined
particle, are solved in the model.
by the contribution of the electric (magnetic) current which
For our study, we have chosen a particle with a quite
determines EDE (MDH ), the electric (magnetic) part of the
complicated shape [49]. In particular, the particle has a special
dipole moment. Similar approaches can be used to explain the
shape inscribed in a sphere and formed by two perpendicularly
scattering features of magnetic particles with different shapes.
arranged circular layers, as shown in the insets of Fig. 5.
In the next section, we use our approach to consider the
Since the modes of such a resonator exist in separate circular
scattering properties of a more complexly shaped magnetic
layers, their polarization stability increases and is dependent
scatterer. In addition, we briefly discuss the use of the COMSOL
on the thickness of the layers. Using layers with different
MULTIPHYSICS software for multipole analysis and compare
thicknesses, it is possible to ensure the detuning of the eigen-
the results of the DDA calculations and COMSOL MULTI-
frequencies of the corresponding modes. In this case, the
PHYSICS.
mode degeneracy inherent in a conventional spherical dielec-
tric particle is lifted.
The wavelength-dependent scattering cross section was
C. DDA versus COMSOL MULTIPHYSICS simulated for the given particle. The simulation was per-
To validate the approach derived above for the calculation formed in the RF module of the COMSOL MULTIPHYSICS
of the exact multipole moments and show the applicability of software (we use the solver Electromagnetic Waves, Fre-
our method, the corresponding procedures are implemented quency Domain in this software). We perform calculations
in the commercial COMSOL MULTIPHYSICS software. This considering that the particle’s material properties are constant
package utilizes the finite-element method to solve the un- (dispersionless and linear) and the fields change sinusoidally
derlying partial differential equations at hand. In COMSOL in time at a known wavelength range. This is the case of
MULTIPHYSICS, the solution is computed by assembling and wavelike field solutions of resonant or radiating structures.
solving a set of equations over all of the finite elements of the In this study, we use the default physics-controlled COMSOL
model. MULTIPHYSICS settings for fine mesh generation. The far-field

155425-7
ANDREY B. EVLYUKHIN AND VLADIMIR R. TUZ PHYSICAL REVIEW B 107, 155425 (2023)

FIG. 5. Scattering cross sections and corresponding multipole decomposition contributions calculated with the use of (a) and (b) the DDA
method and (c) and (d) COMSOL MULTIPHYSICS software for a magnetic-dielectric particle with a radius of 72 mm and constitutive parameters
μ p = 5 and ε p = 5. The thicknesses of the vertical and horizontal circular layers are 25 and 35 mm, respectively. The particle is placed in
an airlike environment (εs = μs = 1) and irradiated by a linearly polarized plane wave. The insets in the top row demonstrate the discrete
presentation of the particle applied for the DDA calculation and the incident wave configurations, whereas the insets in the bottom row show
the patterns of the inner electric and magnetic fields and vectors of the electric p and magnetic m dipole moments calculated in COMSOL
MULTIPHYSICS according to Eqs. (17) at the resonant wavelength λ = 452 mm.

calculations are made on the interior boundary of the PML the particle (“Total”), and (iii) deriving the scattering cross
as the scattering theory suggests. The expressions for the section by summing the dipole and quadrupole contributions
exact multipole moments from Table I with the corresponding (“MULT”).
modification (see above) are implemented in the Component The results of our calculations made using both the DDA
Section of the solver as variables. method and COMSOL MULTIPHYSICS are collected in Fig. 5.
The simulation computes the local electromagnetic field It is easy to see very good agreement between the results,
on each mesh point accompanied by an incident plane wave which confirms the correctness of the obtained expressions.
with the polarization and propagation direction specified, as Nevertheless, while the resonant wavelengths for the scatter-
shown in the insets in Fig. 5. Both frontal and lateral irradia- ing cross section and contributions are the same for these
tion conditions for the incident wave are under consideration. two methods, a slight difference in the amplitude and width
The verification is made by carrying out the calculations in of the resonances is noticed. Additional testing, as expected,
three independent ways: (i) utilizing COMSOL MULTIPHYSICS’s revealed that this difference decreases as the discretization
built-in far-field computation node using the optical theorem increases, although for finer meshes the computation time
(labeled “Opt. Theorem” in Fig. 5), (ii) deriving the scattered rises sufficiently for both methods. We must stipulate here that
energy by integrating the Poynting vector over an imaginary the consistency of the results presented in the figures is deter-
sphere (interior boundary of the PML) constructed around mined primarily by the features of the calculation method used

155425-8
ELECTROMAGNETIC SCATTERING BY ARBITRARY- … PHYSICAL REVIEW B 107, 155425 (2023)

FIG. 6. (a) and (b) Extinction (σext ), scattering (σscat ), and absorption (σabs ) cross sections and (c) and (d) the corresponding contributions
of the electric and magnetic dipole moments for a ferrite sphere irradiated by the left-handed and right-handed circularly polarized waves. The
external static magnetic field H0 is applied in the Faraday configuration (i.e., parallel to the incident radiation propagation, H0 k) as depicted
in the insets in the top row. The insets in the bottom row show the patterns of the inner electric and magnetic fields and vectors of the electric p
and magnetic m dipole moments calculated at the wavelengths of the ED and MD resonances, respectively. The sphere radius is 4.0 mm, and
the wavelength-dependent constitutive parameters of ferrite are given in Appendix B.

in which the obtained expressions for the multipole moments VI. MAGNETIC RESONANCE IN A FERRITE SPHERE
are implemented. Both the DDA and COMSOL MULTIPHYSICS
Finally, in this paper, we apply the obtained expressions
are finite-element methods, whose accuracy depends on the
for multipole moments to study the magnetic circular bire-
density of the mesh partition; however, a comparison of
fringence (Faraday effect) in the scattering of electromagnetic
the accuracies of the calculations by these two methods is
waves by a ferrite sphere. As a material for the sphere, we
beyond the scope of the present work. However, note that
consider yttrium iron garnet (YIG), which is widely used in
since the DDA implementation includes the ordinary practical
practice when constructing nonreciprocal microwave devices
scheme [28,50], the criteria for choosing the discretization
[52]. In the chosen wavelength range, the permittivity ε p of
(the number of dipoles) are the same as described earlier (see
this ferrite is a dispersionless scalar quantity, whereas its
Ref. [51]).
permeability μ p is treated as a nondiagonal tensor acquiring
As an additional benefit, implementing the expressions
dispersion and absorption properties (see Appendix B). They
for multipole contributions in COMSOL MULTIPHYSICS makes
arise as a result of the manifestation of a ferromagnetic reso-
it possible to visualize the orientation of the corresponding
nance caused by an external magnetic bias field. In this study,
vectors [see insets in Figs. 5(c) and 5(d)] and thus facilitate
we consider the direction of the externally applied static mag-
the analysis of the resonant states arising in the complexly
netic field H0 and the wave vector k of the incident wave to
shaped particles.
be coincident, as they should be in the Faraday configuration.

155425-9
ANDREY B. EVLYUKHIN AND VLADIMIR R. TUZ PHYSICAL REVIEW B 107, 155425 (2023)

As is well known, the eigenwaves of an unbounded ferrite we accounted for the fact that the total current density in-
medium propagating along the magnetization direction are the duced in magnetic particles by incoming radiation includes
left-handed (LCP) and right-handed (RCP) circularly polar- the individual contributions from the electric polarization and
ized waves [53], which differ in the propagation constants. magnetization, whereas all multipole moments are divided
From a phenomenological point of view, the Faraday effect into electric and magnetic parts. The final expressions for
is explained by the fact that each of those waves propagates the moments were collected in a tabular form convenient for
in the medium with a different effective permeability μ± = additional software implementations. In particular, we im-
μT ± α, where μ+ and μ− are related to the LCP and RCP plemented them in the calculation procedures of the DDA
waves, respectively [54]. Near the ferromagnetic resonance, method and COMSOL MULTIPHYSICS software. The DDA im-
circular dichroism also takes place, which consists of a sig- plementation was verified by comparing the results with the
nificant attenuation of the LCP eigenwave as it propagates in classical Mie method applied for the calculation of the scat-
the ferrite. The manifestation of these effects can be easily tering cross section of a spherical magnetic particle with
recognized in the field of circularly polarized waves scat- reference to the Kerker effect. The correct fulfillment of
tered by a ferrite sphere as presented in Fig. 6. These data the conditions for the Kerker effect was also verified using
are obtained with the COMSOL MULTIPHYSICS software using disk-shaped particles. For a magnetic particle with a very
equations from Table I for the multipole moments imple- complicated shape, the validation was performed for the DDA
mented inside. implementation versus the COMSOL MULTIPHYSICS software.
In particular, we can see a significant difference in the cross The peculiarities of the manifestation of the Faraday effect
sections for the cases in which the sphere is irradiated either for the circularly polarized waves scattered by a ferrite sphere
by LCP waves or by RCP waves. In the selected wavelength were also discussed.
range, two resonances arise from the contribution of the ED We believe that in the future, the obtained expressions
and MD moments to the total scattering cross section of the could be useful for studying the manifestation of known
LCP wave, whereas in the scattered field of the RCP wave, magneto-optical phenomena (Faraday rotation, Kerr effect,
only an MD resonance appears. The conditions of these reso- etc.) in magnetic particles with a complex shape, as well as
nances are also confirmed by corresponding distributions of for revealing other exotic effects associated with the appear-
the electromagnetic field inside the sphere induced by the ance of toroidal, anapole, and supercavity modes in magnetic
incident wave and the orientation of vectors p and m for these particles and metamaterials in those wavelength ranges where
resonances, which are extracted from COMSOL MULTIPHYSICS. the corresponding constitutive parameters exist.
It is noteworthy that the resonant wavelengths for the MD res-
onance do not coincide for these two orthogonal polarizations;
the MD resonance for the RCP wave is blueshifted compared ACKNOWLEDGMENTS
to its LCP counterpart. Moreover, at the resonant wavelengths, A.B.E. acknowledges funding support from the Deutsche
the absorption cross section of the LCP wave is greater than Forschungsgemeinschaft (DFG, German Research Foun-
that of the RCP wave, which is a manifestation of magnetic dation) under Germany’s Excellence Strategy within the
circular dichroism. Cluster of Excellence PhoenixD (EXC 2122, Project ID No.
An analysis of the multipole contributions to resonant 390833453). V.R.T. is grateful for the hospitality and support
states in a ferrite sphere shows that the multipole moments de- from Jilin University, China.
rived from magnetization M are important when analyzing the
scattering in the presence of an external magnetic bias field.
In particular, it is revealed that although the MD resonance APPENDIX A: EXPRESSIONS OF THE MAGNETIC PARTS
occurs in the scattering spectra of both LCP and RCP waves, Now we show how the nabla operator can be eliminated
the amplitude of the MDH contribution is almost equal to that from the expressions for the magnetic parts of the exact mul-
of the MDE part in the LCP case. tipole moments. Let us consider, as an example, the exact MD
Thus, it follows that for a correct description of scatter- moment:
ing by ferrite particles, the magnetic part of the multipole

moments cannot be ignored. Since the design of dielectric 3 j1 (kr)
metamaterials relies mainly on the multipole decomposition mM = [r × [∇ × M]]dr, (A1)
2 Vp kr
method, accounting for the terms derived from magnetiza-
tion significantly expands the possibilities of the method in where j1 (kr) is the first-order spherical Bessel function. For a
the analysis of structures when considering magneto-optical system including only a single-point MD moment m j located
effects. In particular, this addendum allows one to reveal at the point r j , the magnetization can be written as M(r) =
the separate impacts of the electric and magnetic subsystems m j δ(r − r j ). Substituting this magnetization in Eq. (A1) and
explicitly in the framework of the multipole decomposition performing the integration with the delta function, we can
method. write

VII. CONCLUSIONS 3 j1 (kr)
mM = [r × [∇δ(r − r j ) × m j ]]dr
We derived expressions for the exact multipole moments 2 kr

considering electromagnetic waves scattered by magnetic par- 3 j1 (kr j )
= [r × [∇δ(r − r j ) × m j ]]dr
ticles of an arbitrary shape. When deriving our approach, 2 kr j

155425-10
ELECTROMAGNETIC SCATTERING BY ARBITRARY- … PHYSICAL REVIEW B 107, 155425 (2023)

 
3 j1 (kr j )
− rj × ∇ × mj
2 kr j
j1 (kr j ) 3k 2 j2 (kr j )
=3 mj + [r j × [r j × m j ]], (A2)
kr j 2 (kr j )2
where the following useful relations are used:

[r × [∇δ(r − r j ) × m j ]]dr = 2m j (A3)

and
jn (kr) jn+1 (kr)
∇ = −rk 2 . (A4)
(kr)n (kr)n+1
Replacing m j by Mdr in Eq. (A2) and integrating over the
volume where M = 0, we finally obtain

3 j1 (kr)
mM = [r × [∇ × M]]dr FIG. 7. Wavelength dependences of the relative permeability pa-
2 Vp kr
 rameters of ferrite with the magnetization along the z axis. They are
j1 (kr) related to yttrium iron garnet (G-113) with the following parameters
=3 Mdr
Vp kr (see Appendix H in Ref. [52]): H0 = 3500 Oe, H = 45 Oe, M0 =
 1780/(4π ) G, γ = 1.76086 × 1011 C kg−1 , and scalar permittivity
3k 2 j2 (kr) ε p = 15.
+ [r(r · M) − Mr 2 ]dr. (A5)
2 Vp (kr)2
Using a similar approach, one can consider and transform
the expressions for all multipole moments. The necessary the magnetic losses (the time factor is e−iωt ) [53]:
relations are listed below:
 ⎛ ⎞
μT +iα 0
{r · [∇δ(r − r j ) × m j ]}rdr = [r j × m j ], (A6) ⎝
μ̂ p = −iα μT 0 ⎠, (B1)
 0 0 μL
r 2 [∇δ(r − r j ) × m j ]dr = −2[r j × m j ], (A7)
 where μT = 1 + χ  + iχ  , α =  + i  , and μL = 1,
{r[∇δ(r − r j ) × m j ] + [∇δ(r − r j ) × m j ]r}rdr = 0̂, with χ  = ω0 ωm [ω02 − ω2 (1 − b2 )]D−1 , χ  = ωωm b[ω02 +
(A8) ω2 (1 + b2 )]D−1 , 
= ωωm [ω02 − ω2 (1 + b2 )]D−1 , 
=
 −1
2ω ω0 ωm bD , and D = [ω0 − ω (1 + b )] + 4ω0 ω b ,
2 2 2 2 2 2 2 2

{r · [∇δ(r − r j ) × m j ]}dr = 0, (A9) where ω0 = μ0 γ H0 is the Larmor frequency, ωm = μ0 γ M0

 is the magnetization frequency, and b = μ0 γ H/(2ω) is a
{r · [∇δ(r − r j ) × m j ]}r 2 dr = −(r j · [r j × m j ]), (A10) dimensionless damping constant. The ferrite parameters used
 in our simulations are presented in Fig. 7.
For the given magnetization, the magnetic polarizability
(r · [∇δ(r − r j ) × m j ])rr dr = r j [r j × m j ]+[r j × m j ]r j ,
tensor (26) acquires the form
(A11)

3Vp
r 2 {[∇δ(r − r j ) × m j ]r + r[∇δ(r − r j ) × m j ]}dr α̂ M =
− α2
(ηT )2
⎛ ⎞
= −2([r j × m j ]r j + r j [r j × m j ]). (A12) ξ T ηT − α 2 i3α 0
⎜ −i3α ξ T ηT − α 2 ⎟
×⎝ 0 ⎠,
ξ L
0 0 η L [(η T 2
) − α 2
]
APPENDIX B: CONSTITUTIVE PARAMETERS (B2)
OF FERRITE
We use common expressions for constitutive parameters of where ξ T = μT − 1, ξ L = μL − 1, ηT = 2 + μT , and ηL =
normally (along the z axis) biased ferrite taking into account 2 + μL .

155425-11
ANDREY B. EVLYUKHIN AND VLADIMIR R. TUZ PHYSICAL REVIEW B 107, 155425 (2023)

[1] Q. Zhao, J. Zhou, F. Zhang, and D. Lippens, Mie resonance- in hexagonal arrays of dielectric trimers: Magnetic group ap-
based dielectric metamaterials, Mater. Today 12, 60 (2009). proach, Phys. Rev. B 103, 165402 (2021).
[2] S. Xiao, T. Wang, T. Liu, C. Zhou, X. Jiang, and J. Zhang, [19] A. E. Miroshnichenko, A. B. Evlyukhin, Y. F. Yu, R. M.
Active metamaterials and metadevices: A review, J. Phys. D 53, Bakker, A. Chipouline, A. I. Kuznetsov, B. Luk’yanchuk, B. N.
503002 (2020). Chichkov, and Y. S. Kivshar, Nonradiating anapole modes in
[3] J. Qin, S. Xia, W. Yang, H. Wang, W. Yan, Y. Yang, Z. Wei, dielectric nanoparticles, Nat. Commun. 6, 8069 (2015).
W. Liu, Y. Luo, L. Deng, and L. Bi, Nanophotonic devices [20] E. Zanganeh, A. Evlyukhin, A. Miroshnichenko, M. Song, E.
based on magneto-optical materials: Recent developments and Nenasheva, and P. Kapitanova, Anapole Meta-atoms: Nonra-
applications, Nanophotonics 11, 2639 (2022). diating Electric and Magnetic Sources, Phys. Rev. Lett. 127,
[4] M. G. Barsukova, A. S. Shorokhov, A. I. Musorin, D. N. 096804 (2021).
Neshev, Y. S. Kivshar, and A. A. Fedyanin, Magneto-optical [21] M. V. Rybin, K. L. Koshelev, Z. F. Sadrieva, K. B. Samusev,
response enhanced by Mie resonances in nanoantennas, ACS A. A. Bogdanov, M. F. Limonov, and Y. S. Kivshar, High-Q
Photonics 4, 2390 (2017). Supercavity Modes in Subwavelength Dielectric Resonators,
[5] A. Christofi, Y. Kawaguchi, A. Alù, and A. B. Khanikaev, Giant Phys. Rev. Lett. 119, 243901 (2017).
enhancement of Faraday rotation due to electromagnetically [22] Z. Lin and S. T. Chui, Electromagnetic scattering by optically
induced transparency in all-dielectric magneto-optical metasur- anisotropic magnetic particle, Phys. Rev. E 69, 056614 (2004).
faces, Opt. Lett. 43, 1838 (2018). [23] Y.-L. Geng, X.-B. Wu, L.-W. Li, and B.-R. Guan, Mie scatter-
[6] A. I. Musorin, M. G. Barsukova, A. S. Shorokhov, B. S. ing by a uniaxial anisotropic sphere, Phys. Rev. E 70, 056609
Luk’yanchuk, and A. A. Fedyanin, Manipulating the light inten- (2004).
sity by magnetophotonic metasurfaces, J. Magn. Magn. Mater. [24] B. Zare Rameshti, Y. Cao, and G. E. W. Bauer, Magnetic
459, 165 (2018). spheres in microwave cavities, Phys. Rev. B 91, 214430
[7] K. Bi, J. Zhou, H. Zhao, X. Liu, and C. Lan, Tunable dual- (2015).
band negative refractive index in ferrite-based metamaterials, [25] A. Lakhtakia and T. G. Mackay, Electromagnetic scattering by
Opt. Express 21, 10746 (2013). homogeneous, isotropic, dielectric–magnetic sphere with topo-
[8] K. Bi, W. Zhu, M. Lei, and J. Zhou, Magnetically tunable logically insulating surface states, J. Opt. Soc. Am. B 33, 603
wideband microwave filter using ferrite-based metamaterials, (2016).
Appl. Phys. Lett. 106, 173507 (2015). [26] R. D. Graglia, P. L. E. Uslenghi, and R. S. Zich, Moment
[9] Q. Wang, L. Zeng, M. Lei, and K. Bi, Tunable metamaterial method with isoparametric elements for three-dimensional
bandstop filter based on ferromagnetic resonance, AIP Adv. 5, anisotropic scatterers, Proc. IEEE 77, 750 (1989).
077145 (2015). [27] D. A. Smith and K. L. Stokes, Discrete dipole approximation for
[10] S. Y. Polevoy, S. L. Prosvirnin, S. I. Tarapov, and V. R. Tuz, magnetooptical scattering calculations, Opt. Express 14, 5746
Resonant features of planar Faraday metamaterial with high (2006).
structural symmetry – Study of properties of a 4-fold array of [28] R. A. de la Osa, P. Albella, J. M. Saiz, F. González, and
planar chiral rosettes placed on a ferrite substrate, Eur. Phys. J. F. Moreno, Extended discrete dipole approximation and its
Appl. Phys. 61, 30501 (2013). application to bianisotropic media, Opt. Express 18, 23865
[11] M. Lei, N. Feng, Q. Wang, Y. Hao, S. Huang, and K. Bi, (2010).
Magnetically tunable metamaterial perfect absorber, J. Appl. [29] R. Chen, E. K. N. Yung, Z. Xie, and Y. Wang, FDTD analysis of
Phys. 119, 244504 (2016). EW wave circulating by a magnetized ferrite body in free space,
[12] V. V. Yachin, S. Y. Polevoy, L. I. Ivzhenko, S. I. Tarapov, Int. J. Infrared Millimeter Waves 19, 329 (1998).
and M. I. Nakhimovych, Experimental verification of Faraday [30] H. Liu and H. W. Yang, FDTD analysis of magnetized ferrite
rotation enhancement by all-ferrodielectric metasurface, J. Opt. sphere, J. Electromagn. Waves Appl. 22, 2399 (2008).
Soc. Am. B 36, 261 (2019). [31] V. Schmidt and T. Wriedt, T-matrix method for biaxial
[13] L. I. Ivzhenko, S. Y. Polevoy, S. I. Tarapov, V. V. Yachin, K. anisotropic particles, J. Quantum Spectrosc. Radiat. Transfer
Kurselis, R. Kiyan, and B. N. Chichkov, Experimental observa- 110, 1392 (2009).
tion of tunable Wood type resonances in an all-ferrodielectric [32] J. L.-W. Li, W.-L. Ong, and K. H. R. Zheng, Anisotropic scat-
periodical metasurface, Opt. Lett. 45, 5514 (2020). tering effects of a gyrotropic sphere characterized using the
[14] H. Kettunen, H. Wallén, and A. Sihvola, Cloaking and mag- T-matrix method, Phys. Rev. E 85, 036601 (2012).
nifying using radial anisotropy, J. Appl. Phys. 114, 044110 [33] V. V. Yachin, T. L. Zinenko, and S. V. Mizrakhy, Resonance
(2013). enhancement of Faraday rotation in double-periodic gyromag-
[15] A. Kimel et al., The 2022 magneto-optics roadmap, J. Phys. D netic layers analyzed by the method of integral functionals, J.
55, 463003 (2022). Opt. Soc. Am. B 35, 851 (2018).
[16] V. R. Tuz, V. V. Khardikov, and Y. S. Kivshar, All-dielectric [34] G. P. Zouros, G. D. Kolezas, and I. G. Kyrannas, Complex
resonant metasurfaces with a strong toroidal response, ACS resonances of anisotropic spherical resonators, IEEE Trans.
Photonics 5, 1871 (2018). Antennas Propag. 66, 5282 (2018).
[17] S. Xu, A. Sayanskiy, A. S. Kupriianov, V. R. Tuz, P. [35] C. Qiu, L. Gao, J. Joannopoulos, and M. Soljačić, Light scat-
Kapitanova, H.-B. Sun, W. Han, and Y. S. Kivshar, Experi- tering from anisotropic particles: Propagation, localization, and
mental observation of toroidal dipole modes in all-dielectric nonlinearity, Laser Photonics Rev. 4, 268 (2010).
metasurfaces, Adv. Opt. Mater. 7, 1801166 (2019). [36] P. Grahn, A. Shevchenko, and M. Kaivola, Electromagnetic
[18] V. Dmitriev, S. Domingos Silva Santos, A. B. Evlyukhin, A. S. multipole theory for optical nanomaterials, New J. Phys. 14,
Kupriianov, and V. R. Tuz, Toroidic and antitoroidic orders 093033 (2012).

155425-12
ELECTROMAGNETIC SCATTERING BY ARBITRARY- … PHYSICAL REVIEW B 107, 155425 (2023)

[37] J. G. Van Bladel, in Electromagnetic Fields, IEEE Press Series [45] R. E. Raab and O. L. De Lange, Multipole Theory in Electro-
on Electromagnetic Wave Theory, Vol. 19 (Wiley, Hoboken, NJ, magnetism: Classical, Quantum, and Symmetry Aspects, with
2007). Applications, International Series of Monographs on Physics
[38] A. B. Evlyukhin, C. Reinhardt, and B. N. Chichkov, Mul- Vol. 128 (Oxford University Press, Oxford, 2004).
tipole light scattering by nonspherical nanoparticles in the [46] C. F. Bohren and D. R. Huffman, Absorption and Scattering of
discrete dipole approximation, Phys. Rev. B 84, 235429 Light by Small Particles (Wiley, Hoboken, NJ, 2008).
(2011). [47] A. B. Evlyukhin, C. Reinhardt, A. Seidel, B. S. Luk’yanchuk,
[39] A. B. Evlyukhin, T. Fischer, C. Reinhardt, and B. N. Chichkov, and B. N. Chichkov, Optical response features of Si-
Optical theorem and multipole scattering of light by ar- nanoparticle arrays, Phys. Rev. B 82, 045404 (2010).
bitrarily shaped nanoparticles, Phys. Rev. B 94, 205434 [48] S. Yushanov, J. S. Crompton, and K. C. Koppenhoefer, Mie
(2016). scattering of electromagnetic waves, in Proceedings of the
[40] A. B. Evlyukhin and B. N. Chichkov, Multipole decomposi- COMSOL Conference (Boston, 2013).
tions for directional light scattering, Phys. Rev. B 100, 125415 [49] Y. M. Bezborodov, A. A. Lipatov, T. N. Ustimenko, and V. V.
(2019). Shevchenko, Dielectric resonator, Soviet Union (USSR) patent,
[41] E. A. Gurvitz, K. S. Ladutenko, P. A. Dergachev, A. B. Patent No. SU 1058014 (1982).
Evlyukhin, A. E. Miroshnichenko, and A. S. Shalin, The high- [50] M. A. Yurkin and A. G. Hoekstra, The discrete dipole approx-
order toroidal moments and anapole states in all-dielectric imation: An overview and recent developments, J. Quantum
photonics, Laser Photonics Rev. 13, 1800266 (2019). Spectrosc. Radiat. Transfer 106, 558 (2007).
[42] M. Kerker, D.-S. Wang, and C. L. Giles, Electromagnetic [51] B. T. Draine and P. J. Flatau, Discrete-dipole approximation for
scattering by magnetic spheres, J. Opt. Soc. Am. 73, 765 scattering calculations, J. Opt. Soc. Am. A 11, 1491 (1994).
(1983). [52] D. M. Pozar, Microwave Engineering, 4th ed. (Wiley, Hoboken,
[43] A. B. Evlyukhin, C. Reinhardt, E. Evlyukhin, and B. N. NJ, 2012).
Chichkov, Multipole analysis of light scattering by arbitrary- [53] A. G. Gurevich, Ferrites at Microwave Frequencies (Heywood,
shaped nanoparticles on a plane surface, J. Opt. Soc. Am. B 30, London, 1963).
2589 (2013). [54] V. R. Tuz, M. Y. Vidil, and S. L. Prosvirnin, Polarization
[44] R. Alaee, C. Rockstuhl, and I. Fernandez-Corbaton, An elec- transformations by a magneto-photonic layered structure in
tromagnetic multipole expansion beyond the long-wavelength the vicinity of a ferromagnetic resonance, J. Opt. 12, 095102
approximation, Opt. Commun. 407, 17 (2018). (2010).

155425-13