It is known that the two eigen plane waves incident to the generalized soft-and-hard/DB (GSHDB) boundary are reflected as from the PEC or PMC boundary, i.e., with reflection coefficients −1 or +1, for any angle of incidence. The present paper discusses a more general class of boundaries by requiring that the reflection coefficients R+ and R−, corresponding to the two eigen plane waves, have opposite values, R±=±R with R independent of the angle of incidence. It turns out that, there are two possibilities, R=1 for the class of GSHDB boundaries, and R=j for another class, extending that of the perfect electromagnetic conductor (PEMC) boundaries. Matched waves at, and plane-waves reflected from, boundaries of the latter class are studied in the paper.
Introduction
Boundary conditions are fundamental building blocks in problems of mathematical physics, including electromagnetics. For given sources in a given space, Maxwell equations require additional constraints on the boundaries of the domain that force uniqueness and existence of the solution. These conditions between the electric and magnetic fields and fluxes at the boundary can in principle be rather general [1]; however, in finite real-world problems, the boundaries often are characterized by approximate surface conditions [2, 3].
Examples of such boundary conditions are perfectly electrically conducting (PEC) surfaces, artificial magnetic conductor and high-impedance surfaces [4, 5], and anisotropic soft-and-hard surfaces [6]. But the possibilities to engineer the electromagnetic response of complex boundaries have a great variety. Thin two-dimensional structures which are capable of modifying the reflection and transmission of the incident wave in multiple manners have been recently designed and fabricated, and the label metasurface has appeared in the electromagnetics literature to cover such advanced surfaces [7]. Generalizing the traditional frequency-selective surfaces and transmit/reflect arrays [8, 9], these metasurfaces can control the amplitude and phase of the wavefront in various ways [10], the effects can be nonreciprocal [11], non-local, or even active [12]. In the optical range of the electromagnetic spectrum, these developments within the metasurface paradigm are known by labels like planar photonics [13] and flat optics [14].
It is the purpose of the present paper to investigate still uncharted classes of complex electromagnetic boundaries and surfaces. The treatment is purely theoretical, concentrating on certain aspects of electromagnetic boundary conditions beyond those in the book [1] by these authors. Theoretical predictions of unconventional boundaries, like, as example, the perfect electromagnetic conductor PEMC [15] have found experimental realizations [16], and found practical applications in the manipulation of properties of electromagnetic waves. The novel class of boundary conditions introduced in the present paper may be of use in the future design of novel metaboundaries.
General Boundary Conditions
A typical electromagnetic problem involves solving the Maxwell equations in a region bounded by a surface with imposed boundary conditions. The most general (linear and local) boundary conditions (GBC) can be shown to take the form [1, 17, 18],
α1cn⋅B+ϵ0β1n⋅D+a1t⋅E+η0b1t⋅H=0,(1)
α2cn⋅B+ϵ0β2n⋅D+a2t⋅E+η0b2t⋅H=0,(2)
with
c=μ0ϵ01,η0=ϵ0μ0.(3)
The unit vector n is normal, and the vectors a1t⋯b2t are tangential, to the boundary surface. They are dimensionless and so are the scalars α1,α2,β1,β2. Because one of the scalar coefficients in each equation can be chosen, the number of free parameters defining the general boundary conditions is 10. Realization of a boundary defined by general conditions of the form (1) and (2), in terms of the interface of a bianisotropic medium, has been discussed in [1], Section 5.14.
In the present paper we assume for simplicity that the boundary surface n⋅r=0 is planar, as defined by a constant unit vector n. Also, we assume that the medium is isotropic with parameters ϵ0 and μ0, in which case the boundary conditions (1) and (2) can be rewritten in the simpler equivalent form [1, 18]
a1⋅E+b1⋅η0H=0,(4)
a2⋅E+b2⋅η0H=0.(5)
where the four vectors a1⋯b2 may contain components normal and tangential to the boundary,
ai=ainn+ait,bi=binn+bit,i=1,2.(6)
It is assumed that (4) and (5) represent two linearly independent conditions, because, otherwise, they do not define a unique boundary-value problem.
Special Cases
The boundary conditions (4) and (5) contain a lot of familiar special cases starting from the perfect electric conductor (PEC), defined by
a1t⋅E=a2t⋅E=0,a1t×a2t=0,(7)
and the perfect magnetic conductor (PMC), defined by
b1t⋅H=b2t⋅H=0,b1t×b2t=0.(8)
A straightforward generalization of these two conditions is that of the perfect electromagnetic conductor (PEMC) [15, 19, 20],
b1t⋅(H+ME)=b2t⋅(H+ME)=0,(9)
or, in vector form,
n×(H+ME)=0,(10)
where M is the PEMC admittance.
As other special cases of the GBC conditions (1) and (2) we may add
contain SH and DB conditions (12), (11) as special cases.
The generalized soft-and-hard/DB (GSHDB) conditions [25],
at⋅E+αn⋅ηoH=0,βn⋅E+bt⋅ηoH=0,(16)
represent a further generalization of (15).
The impedance conditions
a1t⋅E+b1t⋅ηoH=0,a2t⋅E+b2t⋅ηoH=0.(17)
Purpose of this Study
In a previous study [25], it has been shown that a boundary defined by the GSHDB conditions (16) has the following property: any plane wave incident from an isotropic half space to the boundary can be decomposed in two eigen plane waves, one of which is reflected as from the PEC boundary, and, the other one, as from the PMC boundary. Since the GSHDB conditions (16) contain (11)–(13) and (15) as special cases, the same property is shared by all the corresponding boundaries. The converse property was shown in [17, 1]: if the PEC/PMC decomposition property is required to be valid for the eigenwaves associated to the GBC conditions (4), (5) (reflection coefficient either +1 or −1), the boundary must actually satisfy the GSHDB conditions (16). Thus, the generalized soft-and-hard/DB boundary is the most general boundary with the PEC/PMC equivalence property.
It is the purpose of the present paper to extend this theory by requiring that the reflection coefficients R+,R− corresponding to the two eigen plane waves satisfy R±=±R, where R is independent of the angle of incidence (k vector of the incident wave). For the special case of the GSHDB boundary we have R=1.
Reflection of Eigen Plane Waves
Let us consider time-harmonic plane waves satisfying the plane-wave equations
k×E=koηoH,(18)
k×ηoH=−koE.(19)
and denote fields incident to, and reflecting from, the boundary surface, respectively by
Figure 1. Plane wave incident on a surface characterized by a general boundary condition GBC. The incident and reflected wave vectors ki,kr are decomposed into their tangential and normal components.
ki=kt−knn,kr=kt+knn,(21)
ki⋅ki=kr⋅kr=ko2=ω2μoϵo.(22)
Eigen plane waves are defined by requiring that the tangential components of the electric fields at the boundary satisfy a relation of the form
Etr=REti,(23)
i.e., that the incident and reflected fields tangential to the boundary have similar polarization and are related by of a scalar factor R, the reflection coefficient. Polarizations and reflection coefficients of the eigen plane waves depend on the vector parameters of the GBC boundary conditions (4), (5). Applying a previous analysis [1, 17], the reflection coefficient can be shown to satisfy the condition
for i=1,2, appears useful in the subsequent analysis.
In the general case, the solutions R of the equation (24) depend on the boundary parameters a1⋯b2 and the tangential component of the wave vector, kt. Requiring that the solutions R be independent of the wave vector sets conditions for the boundary parameter vectors. For example, for the GSHDB boundary, (24) has the solutions R=+1 and R=−1 for any kt[17].
Requiring this to be valid for any angle of incidence, i.e., for any tangential vector kt, leads to the two conditions
n⋅(a1×a2−b1×b2)=0,(46)
n×(a1×b2−a2×b1)=0.(47)
They are respectively equivalent to
a1t×a2t−b1t×b2t=0,(48)
b2na1t−b1na2t+a2nb1t−a1nb2t=0.(49)
For the GSHDB boundary (16), these are identically satisfied.
With (44), the condition (33) becomes
C2R2+C0=0,(50)
whence
R±=±−C0/C2.(51)
Requiring that the solution R of (50) be independent of kt, (50) can be split in the following set of equations relating the boundary parameters a1⋯b2 and the reflection coefficient R:
Derivation of this set of equations is given in the Appendix.
Now we can separate three possible cases:
R2=1
R2=−1
R2=1 and R2=−1.
In the last case, all expressions in (52)–(56) multiplying (R2−1) or (R2+1) must vanish. Actually, this corresponds to setting C0=C1=C2=0 in (33), which means that R may have any value. Ignoring this, we have only two possibilities, either R2=1, or R2=−1. Each of these defines a class of boundaries through certain conditions for the vector parameters a1⋯b2 arising from (52)–(56). Since it has been previously shown [1, 17] that the case R2=1 corresponds to the class of GSHDB boundaries defined by (16), let us concentrate on the case R2=−1, which yields a second possibility for the class of boundaries satisfying R+=−R−. In this case, the reflection coefficients for the two eigenwaves are +j and −j.
Boundaries Defined by R+=−R−=j
From (52), (55) and (56) we obtain conditions for the boundary corresponding to the case R2=−1:
a1na2t−a2na1t+b1nb2t−b2nb1t=0(57)
b1na2n−b2na1n=0(58)
b1ta2t+a2tb1t−b2ta1t−a1tb2t=0(59)
which, together with (48) and (49), define the present class of boundaries.
It has been shown that the case R2=1 corresponds to the class of GSHDB boundaries, for which the vector parameters satisfy n⋅a1×a2=0 and n⋅b1×b2=0, as is seen from (16). Similarly, from (48) and (54) we can conclude that the case R2=−1 corresponds to n⋅a1×a2=0 and n⋅b1×b2=0. Because of this, we can assume that the tangential vectors b1t,b2t and a1t,a2t form two linearly independent pairs, whence we can expand
(b1tb2t)=B(a1ta2t)(60)
for some matrix
B=(B11B21B12B22)(61)
defined by four scalars B11⋯B22. The condition (48) requires
detB=B11B22−B12B21=1.(62)
From (49) we obtain a relation between the normal components of the four vectors as
From (64) and (65) we obtain the corresponding possibilities,
b1=a1t−na1n,b2=a2t−na2n,(72)
and
b1=−a1t+na1n,b2=−a2t+na2n.(73)
(57) and (58) are now satisfied for any given vectors a1 and a2. Substituting (72) or (73), the conditions (46), (47) and (52)–(56) are also satisfied.
Thus, the above analysis offers two possibilities for the boundary conditions corresponding to the case R2=−1,
a1t⋅(E+ηoH)+a1nn⋅(E−ηoH)=0,(74)
a2t⋅(E+ηoH)+a2nn⋅(E−ηoH)=0,(75)
and
a1t⋅(E−ηoH)+a1nn⋅(E+ηoH)=0,(76)
a2t⋅(E−ηoH)+a2nn⋅(E+ηoH)=0.(77)
The conditions (74), (75) can be expressed in vector form as
In the present case R2=−1 we have assumed An=0. In the special case of impedance boundaries, the boundary parameters satisfy a1n=a2n=0, which corresponds to pt=0. In this case, the condition (78) becomes
n×(E+ηoH)=0,(82)
and, the condition (79) becomes
n×(E−ηoH)=0.(83)
(82) and (83) equal special cases of the PEMC boundary conditions (9), with admittances M=1/ηo and M=−1/ηo, respectively. The reflection coefficients for the two eigenwaves at the general PEMC boundary are known to have the form [1], Eq. (2.38),
R±=1∓jMηo1±jMηo,(84)
which satisfies R±2=−1 for both Mηo=1 and Mηo=−1.
One can easily verify that, for both of the boundary conditions (78) and (79), the two solutions for the reflection coefficient equation (33) satisfy both R+=−R− and R±2=−1. The details are given in the Appendix.
In conclusion, requiring that the reflection coefficients R+ and R− of the two eigenwaves satisfy the condition R+=−R−=R, with R2=−1 for any angle of incidence, the boundary is defined either by the conditions (78) or (79).
Extended PEMC Boundaries
The conditions (78) and (79) are special cases of the vector condition
n×(ME+H)+ptn⋅(ME−H)=0,(85)
defining what can be called the class of Extended PEMC (EPEMC) boundaries. The class depends on two parameters, a scalar M and a vector pt. The boundary vector parameters corresponding to (85) are
b1=(a1t−na1n)/Mηo,b2=(a2t−na2n)/Mηo.(86)
For pt=0, the condition of the EPEMC boundary (85) is reduced to that of the PEMC boundary (9), in which case the parameter M equals the admittance of the PEMC boundary.
Because (78) and (79) are special cases of (85), respectively corresponding to M=1/ηo and M=−1/ηo, they can be called conditions of the EPEMC+ and EPEMC− boundaries, respectively. For pt=0 both of them are reduced to the corresponding special PEMC boundaries.
It has been previously shown that a plane wave incident normally to the PEMC boundary is reflected cross polarized, i.e., satisfying Etr⋅Eti=0, exactly when the PEMC admittance satisfies either Mηo=+1 or =−1[1]. It will turn out that the same property is valid for the EPEMC boundary.
Matched Waves
By definition, a plane wave is matched to a boundary when the boundary conditions are identically satisfied for the single plane wave, i.e., there is no reflected wave [1]. Applying the plane-wave equations (18) and (19), the conditions of the EPEMC+ and EPEMC− boundaries, (78) and (79), can be respectively expanded as
Omitting the case when k satisfies the dispersion equation (92) of the matched wave, the dyadic in (106) has a two-dimensional inverse. In this case, the fields at the boundary must satisfy
n×(Ei+Er+ηo(Hi+Hr))=0,(107)
for any vector pt. Actually, (107) equals the boundary condition of the PEMC boundary (10), for M=1/ηo.
The EPEMC− boundary can be handled similarly by starting from the condition (79), leading to
for any vector pt. Again, this equals the boundary condition of the PEMC boundary (10), for M=−1/ηo.
To summarize, the class of boundaries defined by the property R2=−1 consists of two subclasses, EPEMC+ and EPEMC−, which are equivalent in reflection to two subclasses of PEMC boundaries, respectively defined by M=1/ηo and M=−1/ηo, for any vector pt.
As a example, let us consider a plane wave with normal incidence, kr=−ki=nko, reflecting from an EPEMC boundary. The fields are tangential to the boundary: n⋅Ei=n⋅Er=0, n⋅Hi=n⋅Hr=0. From (85), the reflected field can be expressed as
Er=−M2ηo2+11((M2ηo2−1)It+2Mηon×I)⋅Ei.(110)
whence
Er⋅Ei=C(Ei⋅Ei)(111)
where C, defined by
C=−M2ηo2+1M2ηo2−1,(112)
is a measure of copolarization in the reflected field. This function is depicted in Figure 2.
Figure 2. Visualization of the co-polarization quantity C of (112), as a function of the parameter Mηo, for a normally incident plane wave reflecting from an EPEMC boundary defined by (85). The reflected field is co-polarized for Mηo=0 (PMC) and Mηo→±∞ (PEC) while, for Mηo=1 (EPEMC+) and Mηo=−1 (EPEMC−), the reflected field is cross polarized, C=0.
The reflected field appears cross polarized when Er⋅Ei=0, or C=0. This happens for Mηo=±1, i.e., when the EPEMC boundary is either EPEMC+ or EPEMC−. The same property is known to exist for PEMC boundaries with M=1/ηo or M=−1/ηo ([1], Section 2.4). For M=0 (PMC, C=1) and M→±∞ (PEC, C=−1), the reflected field is totally copolarized. Note that, for a circularly polarized incident wave, satisfying Ei⋅Ei=0, from (110) we have Er⋅Ei=0 for any M=±j/ηo.
Conclusion
The problem considered in this paper has been to generalize the property of the class of electromagnetic boundaries known as that of generalized soft-and hard/DB (GSHDB) boundaries. The class of boundaries is known to possess the unique property that the two eigen plane-waves associated to the boundary are reflected with reflection coefficients R+=1 and R−=−1 for any angle of incidence. Such a property has been called PEC/PMC equivalence, because the two eigenwaves are reflected as from a PEC or a PMC boundary. In the present case, the possibility of a class of boundaries with a more general property was studied by requesting that the reflection coefficients of the two eigenwaves satisfy a condition of the form R±=±R, where R is independent of the angle of incidence.
It was shown that, under this condition, R must be either 1 or j. While the former case corresponds to the class of GSHDB boundaries, the latter case was shown to correspond to another class of boundaries defined by two possible sets of boundary conditions, (78) and (79). Both of these were shown to be special cases of a more general class of Extended Perfect Electromagnetic Conductor (EPEMC) boundaries defined by conditions of the form (85), depending on a scalar M and a vector pt tangential to the boundary surface. The two special cases of (85) were respectively called EPEMC+ and EPEMC− boundaries as they correspond to the special parameter values M=1/ηo and M=−1/ηo, respectively. For pt=0, the EPEMC boundary is reduced to the PEMC boundary with M equal to the PEMC admittance. Matched waves and plane-wave reflection properties were studied for both EPEMC+ and EPEMC− classes of boundaries. While the matched waves depend on the pt vector, reflection of the eigenwaves does not. Actually, in reflection, the EPEMC+ and EPEMC− classes of boundaries are equivalent to the corresponding classes of PEMC boundaries defined by the respective admittance values M=1/ηo and M=−1/ηo, for any vector pt.
Appendix
Expanding A and B
Applying (32), we can expand the expressions (39) and (40) as
For (120) to be satisfied for any kt, parts odd and even in kt must vanish separately. This leads to the respective conditions
(R2−1)Ko+(R2+1)Lo=0,(125)
(R2−1)Ke+(R2+1)Le=0.(126)
As a special case, for the GSHDB boundary (16) with a1=at, b1=αn, a2=βn and b2=bt, we find Lo=0 and Le=0. Since at and bt may be any tangential vectors and α,β any scalars, Ko and Ke are nonzero. Thus, both (125) and (126) yield R2=1, or R±=±1, a result known from [1, 25].
Let us consider the general case. For (125) to be valid for any polarization of kt, we must have
These two conditions correspond to the equation (125). Similarly, (126) can be split in three conditions, based on different orders of magnitude in kt, as
Since the trace of (136) covers the condition (130), the latter can be omitted.
In conclusion, the medium parameters a1⋯b2 and the reflection coefficient R must satisfy the conditions (128), (129), (131), (134) and (136)
Expanding C1=0
Let us find the conditions for the boundary parameters a1⋯b2 in the case when the coefficient C1 vanishes for all wave vectors kt. Because the second term of (42) changes sign in kt→kt, from C1=0 we obtain the two conditions
n⋅(a1×a2−b1×b2)=0,(137)
(a1×b2+b1×a2)×n=0.(138)
Assuming n⋅a1×a2=0, the tangential vectors a1t,a2t make a planar basis and we can expand
In conclusion, vectors b1 and b2, depending through (139) and (142) on any vectors a1 and a2 whose tangential components are linearly independent, in terms of any parameters B11⋯B22 satisfying (140), yield C1=0 for any kt.
Verifying EPEMC+ and EPEMC− boundary conditions
Let us verify that the EPEMC+ boundary conditions (74) and (75), and the EPEMC− boundary conditions (76) and (77), depending on the two vectors a1 and a2, both correspond to the property R±2=−1.
which, from (37), ensures the validity of the condition R+=−R−.
Further, we can expand
a1×a2+b1×b2=2Ann,(148)
b1a2−b2a1=∓An(n×iˉ+npt+ptn),(149)
with the upper sign corresponding to EPEMC+, and the lower sign to EPEMC− boundary conditions. Substituting these in (41) and (43) yields
C0=kokn−2n⋅kiAn(ko±pt⋅kt),(150)
C2=kokn2n⋅krAn(ko±pt⋅kt)=C0,(151)
whence, from (37), the reflection coefficients are j and −j for both of the EPEMC+ and EPEMC− conditions.
References
[1]Lindell, I.V. and A. Sihvola, Boundary Conditions in Electromagnetics, Hoboken, N.J., Wiley and IEEE Press, 2020.DOI
[2]Hoppe D.J. and Y. Rahmat-Samii, Impedance Boundary Conditions in Electromagnetics, Washington, DC, USA: Taylor & Francis, 1995.
[3]Senior T.B.A. and J.L. Volakis, Approximate Boundary Conditions in Electromagnetics, London, U.K.: IEE, 1995.
[4]Sievenpiper, D., Lijun Zhang, R.F.J. Broas, N.G. Alexopolous, and E. Yablonovitch, ”High-impedance electromagnetic surfaces with a forbidden frequency band,” IEEE Transactions on Microwave Theory and Techniques, vol. 47, no. 11, pp. 2059-2074, Nov. 1999.DOI
[5]Fong, B.H., J.S. Colburn, J.J. Ottusch, J.L. Visher, and D.F. Sievenpiper, ”Scalar and tensor holographic artificial impedance surfaces,” IEEE Trans. Antennas Propag., vol. 58, no.10, pp. 3212-3221, Oct. 2010.DOI
[6]Kildal, P.-S., ”Artificially soft and hard surfaces in electromagnetics”, IEEE Trans. Antennas Propagat., vol. 38, no. 10, pp. 1537-1544, October 1990.DOI
[7]Holloway C.L., E.F. Kuester, J.A. Gordon, J. O’Hara, J. Booth, and D.R. Smith, ”An overview of the theory and applications of metasurfaces: the two-dimensional equivalents of metamaterials”, IEEE Antennas Propag. Mag., vol. 54, no. 2, pp. 10-35, April 2012.DOI
[8]Munk B., Frequency Selective Surfaces: Theory and Design, (New York: Wiley), 2000.
[9]Maci S. and A. Cucini, ”FSS-based EBG metasurfaces”, in Metamaterials: Physics and Engineering Explorations, ed., N. Engheta and R. Ziolkowski (Piscataway, NJ: IEEE), 2006.
[10]Gok G. and A. Grbic, ”Tailoring the phase and power flow of electromagnetic fields, Phys. Rev. Lett., vol. 111 233904, 2013.pubmed.ncbi.nlm.nih.gov/24476273
[11]Lavigne G. and C. Caloz, ”Magnetless reflective gyrotropic spatial isolator metasurface”, New Journal of Physics, vol. 23, no.7, pp. 1-11, July 2021.DOI
[12]Monticone F., C.A. Valagiannopoulos, and A. Alù, ”Aberration-free imaging based on parity-time symmetric nonlocal metasurfaces”, Phys. Rev. X, vol. 6, 041018, 2016.arxiv.org/abs/1509.07300
[13]Kildishev A.V., A. Boltasseva, and V.M. Shalaev, ”Planar photonics with metasurfaces” Science, vol. 339, 1232009, 2013.DOI
[14]Yu N. and F. Capasso, ”Flat optics with designer metasurfaces”, Nature Mater., vol. 13, 139–50, 2014.DOI
[15]Lindell, I. V. and A. Sihvola, “Perfect electromagnetic conductor”, J. Electro- Waves Appl. Vol. 19, No. 7, 861–869, 2005.DOI
[16]Shahvarpour, A., T. Kodera, A. Parsa, and C. Caloz, ”Arbitrary electromagnetic conductor boundaries using Faraday rotation in a grounded ferrite slab” IEEE Trans. Microwave Theory Tech., vol. 58, no. 11, pp. 2781-2793, 2010.DOI
[17]Lindell, I. V. and A. Sihvola, ”Electromagnetic boundaries with PEC/PMC equivalence,” Prog. Electromag. Res. Lett., Vol.61, pp.119–123, 2016.arxiv.org/abs/1608.05519
[18]Lindell, I. V. and A. Sihvola, ”Electromagnetic wave reflection from boundaries defined by general linear and local conditions,” IEEE Trans. Antennas Propag., vol.65, No.9, 4656–4663, 2017.DOI
[19]Lindell, I. V. and A. Sihvola, ”Transformation method for problems involving perfect electromagnetic conductor (PEMC) structures,” IEEE Trans. Antennas Propag., vol.53, No.9, 3012–3018, 2005.DOI
[20]Lindell, I. V. and A. Sihvola, “Generalization of perfect electromagnetic conductor boundary”, IEEE Trans. Antennas Propag. Vol. 68, No. 11, 7406–7413, 2020.DOI
[21]Lindell, I. V. and A. Sihvola, “Electromagnetic boundary condition and its realization with anisotropic metamaterial,” Phys. Rev. E, Vol. 79, No. 2, 026604 (7 pages), 2009.pubmed.ncbi.nlm.nih.gov/19391860
[22]Kildal, P.-S., “Definition of artificially soft and hard surfaces for electromagnetic waves,” Electron. Lett., Vol. 24, pp. 168–170, 1988.DOI
[23]Lindell, I. V., “Generalized soft-and-hard surface,” IEEE Trans. Antennas Propag., Vol. 50, No. 7, pp. 926–929, July 2002.DOI
[24]Lindell, I. V. and A. Sihvola, “Soft-and-hard/DB boundary conditions realized by a skewon-axion medium,” Trans. IEEE Antennas Propag., Vol. 61, No. 2, pp. 768–774, 2013.DOI
[25]Lindell, I. V. and A. Sihvola, “Generalized Soft-and-hard/DB boundary,” Trans. IEEE Antennas Propag., Vol. 65, No.1, pp. 226–233, 2017.DOI
[26]I.V. Lindell, Methods in Electromagnetic Field Analysis,” 2nd ed., New York: Wiley, 1995.