You are on page 1of 18

Waves in Random and Complex Media

ISSN: 1745-5030 (Print) 1745-5049 (Online) Journal homepage: http://www.tandfonline.com/loi/twrm20

An electromagnetic parameters extraction method


for metamaterials based on phase unwrapping
technique

Yan Shi, Zhi-Yi Li, Long Li & Chang-Hong Liang

To cite this article: Yan Shi, Zhi-Yi Li, Long Li & Chang-Hong Liang (2016): An electromagnetic
parameters extraction method for metamaterials based on phase unwrapping technique,
Waves in Random and Complex Media, DOI: 10.1080/17455030.2016.1165899

To link to this article: http://dx.doi.org/10.1080/17455030.2016.1165899

Published online: 07 Apr 2016.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=twrm20

Download by: [Library Services City University London] Date: 12 April 2016, At: 09:17
Waves in Random and Complex Media, 2016
http://dx.doi.org/10.1080/17455030.2016.1165899

An electromagnetic parameters extraction method for


metamaterials based on phase unwrapping technique
Yan Shi, Zhi-Yi Li, Long Li and Chang-Hong Liang
School of Electronic Engineering, Xidian University, Xi’an, China
Downloaded by [Library Services City University London] at 09:17 12 April 2016

ABSTRACT ARTICLE HISTORY


A new method of extracting effective permittivity and permeability of Received 6 August 2015
metamaterials has been developed in this paper. Scattering parameters Accepted 9 March 2016
are used to represent effective wave impedance and refractive index
of the metamaterials. Considering analytical continuation of effective
refractive index of the metamaterials in the upper half of a complex
angular frequency plane, a phase unwrapping technique has been
implemented to solve branch ambiguity of the refractive index
caused by complex inverse cosine function. An empirical formula to
estimate the starting frequency in the phase unwrapping technique
has been developed to improve the efficiency of the parameter
retrieval procedure. Numerical results including five slabs of rods
and split-ring resonators (SRRs) and a structure composed of two
SRRs and four capacitively loaded strips are given to demonstrate
good accuracy, high efficiency, and noise insensitivity of the proposed
retrieval method.

1. Introduction
In recent years, much attention has been given to electromagnetic metamaterials [1] due to
their unusual electromagnetic properties not found in natural media. A tremendous amount
of work [2,3] has been done to demonstrate novel properties of the metamaterials including
reversal of both Doppler shift and Cherenkov radiation, negative refraction, negative group
velocity, etc.
The design of the metamaterials requires first estimating their effective properties.[4]
Roughly speaking, most of the reported extraction approaches of the effective parameters
involving the wave impedance, index of refraction, permittivity, and permeability are based
on either simulated or measured scattering parameters. Smith et al. used the relationship
between S parameter and T parameter to extract the effective electromagnetic parameters
of single layer structure consisting of the split ring resonators (SRRs) and rods.[5,6] However,
this method is only reasonable when the metamaterial structure is very electrically thin.
With the increase in the thickness of the metamaterial structure, the extraction method
becomes problematic due to multiple values of the refraction index at some frequencies,
which is called branch problem.

CONTACT  Yan Shi  shiyan@mail.xidian.edu.cn


© 2016 Informa UK Limited, trading as Taylor & Francis Group
2    Y. Shi et al.

In order to solve the branch problem, various improved methods have been developed. In
[7], the complex exponential function is expanded in terms of Taylor’s series and the branch
index is iteratively determined. However, the requirement of positive imaginary parts both
of the permittivity and permeability cannot be satisfied at some resonant frequencies, and
thus the method becomes invalid at these corresponding frequencies. Based on the fact that
real part of the refraction index needs to handle the branch problem, while imaginary part
of the refraction index is independent of the branch problem, a retrieval method based on
Kramers–Kronig (K–K) relationship has been proposed.[8] With the K–K relationship, the real
part of the refraction index can be approximately obtained when the imaginary part of the
refraction index is known. But it is not clear how to estimate errors when the integral in the
K–K relationship is truncated. In addition, the branch problem can be solved by measuring
Downloaded by [Library Services City University London] at 09:17 12 April 2016

the metamaterial slabs with different lengths twice,[9] which doubles measurement costs
of the metamaterials. Alternatively, phase unwrapping technique can correctly choose the
branch by detecting successive phase jumps that exceed π at specific frequencies. For an
isotropic material sampling, the phase unwrapping technique generally uses both S11 and S21
to extract its electromagnetic parameters.[10–15] If only S11 is used in the phase unwrapping
technique, the material sampling has to be measured twice.[16] With the phase unwrapping
technique, abrupt phase jumps of metamaterials near the resonant frequency region can
be corrected.[14] Compared with the retrieval method based on K–K relationship, the phase
unwrapping technique is simple and only requires a limited frequency spectrum.[17,18]
Starting frequency of the phase unwrapping technique is generally chosen as zero frequency
because its phase origin corresponds to zero frequency.[10–18] However, efficiency of the
resulting retrieval procedure will be greatly decreased, when desirable frequency band is
far higher than low frequency.
In this paper, the phase unwrapping technique is used to extract the effective electromag-
netic parameters of metamaterials. An estimated formula to estimate the starting frequency
of the phase unwrapping technique has been developed to improve efficiency of parameter
retrieval procedure. Noise sensitivity analyses of the phase unwrapping technique including
magnitude and phase of the S parameters are discussed. Numerical results demonstrate that
the proposed phase unwrapping technique is highly efficient, accurate, and insensitive to
magnitude noise and small phase noise of the S parameters.

2.  Phase unwapping technique to extract electromagnetic parameters


Consider a plane wave with a normal incidence on a metamaterial slab with the thickness
of d. According to the effective medium theory, the metamaterial slab can be characterized
by the effective material parameters. In this paper, the time harmonic e−iωt convention is
used.√The relationship between the S parameters, the normalized effective wave impedance

z = 𝜇∕ 𝜀 and the effective refraction index n = 𝜇𝜀 can be expressed as below:
R(1−ei2nk0 d )
S11 = 1−R2 ei2nk0 d
(1−R2 )ei2nk0 d (1)
S21 = 1−R2 ei2nk0 d

where R = (z − 1)/(z + 1) and k0 is the wave number in free space, and ɛ and μ denote the
relative permittivity and permeability, respectively. By solving (1), we can obtain
Waves in Random and Complex Media   3

√ 2
(1+S11 ) −S212
z=± 2
(1−S11 ) −S212 (2)
S21
eink0 d = 1−S R .
11

The sign in the first equation of (2) can be determined by requiring one of two following
conditions [8]: one is Re(z) > 0 and Im(n) > 0; the other is |ei2nk0 d | < 1. According to the second
equation of (2), the refraction index n can be solved as follows:
[ ( ) ]
1 S21
n= ln ± 2im𝜋 (3)
ik0 d 1 − S11 R
Downloaded by [Library Services City University London] at 09:17 12 April 2016

in which the branch index m caused by the complex logarithmic function is an integer. The
branch index m is equal to zero for the metamaterial slab with electrically small thickness,
while with the increase in the thickness of the metamaterial slabs m is not always zero.
Further (3) can be rewritten as
[ ( ) ( )]
1 S21 S21
n= Im ln − 2m𝜋 − i ⋅ Re ln . (4)
k0 d 1 − S11 R 1 − S11 R

It can be seen from (4) that the branch index m is only related to the real part of the refrac-
tion index n.
In order to determine the branch index m, the second equation of (2) can be reformu-
lated as

eink0 d = Aei(𝜙+2m𝜋) (5)


in which

| S |
| 21 |
A=| |
| 1 − S11 R | (6)
| |
( )
S21
𝜙 = arg (7)
1 − S11 R

where arg (⋅) is called the principal value of the argument of the complex function. In this
scenario, we have ϕ ∊ (−π, π]. Further, we express the refraction index n as

n = n� + in�� . (8)
Substituting (8) into (5), we can get
ln A
n�� = − (9)
k0 d

𝜙 + 2m𝜋
n� = . (10)
k0 d
4    Y. Shi et al.

The analytical continuation of the refractive index n in the upper half of a complex angular
frequency plane is satisfied [19], when the electrical size of the unit cell is smaller than the
operating wavelength. Therefore, the phase ϕ + 2 mπ is continuous. Here, a phase unwrap-
ping technique can be used to determine the branch index m. Unwrapping means connect-
ing the discontinuous −π-to-π segments into a continuous curve. The real part of refractive
index n′ and the phase ϕ approach zero, when frequency decreases to zero. In this scenario,
we have m = 0. Therefore, zero frequency is chosen as the frequency corresponding to phase
origin. With the increase in the frequency, if ϕ = π at one frequency point f0 and ϕ = −π at
next frequency point f1, we have m = 1 to maintain the continuous phase of the refractive
index. On the other hand, if ϕ = −π at one frequency point f0 and ϕ = π at next frequency
point f1, m = −1 is obtained. This procedure can be iteratively implemented across the con-
Downloaded by [Library Services City University London] at 09:17 12 April 2016

cerned frequency band. Once the refractive index and wave impedance are obtained, the
permittivity and permeability can be calculated as follows:
n
𝜀= (11)
z

𝜇 = nz. (12)
It is worthwhile pointing out that the starting frequency of the phase unwrapping technique
is generally set as zero frequency, because the phase origin corresponds to zero frequency.
However, when the desirable frequency band is far higher than low frequency, the phase
unwrapping technique becomes time-consuming due to plenty of the scattering parameter
simulations out of the concerned frequency band. This problem becomes more serious for
multiple slabs of the metamaterial structures due to the use of more memory and computa-
tional time. In fact, the starting frequency in the phase unwrapping technique doesn’t have
to be zero frequency, which is different from the K–K algorithm, because lower and upper
limits of frequency in the K–K integral must be zero and infinite, respectively [19]. It can be
found that a frequency can be chosen as the starting frequency only if the branch index m
is equal to zero at the corresponding frequency. In this scenario, the starting frequency in
the phase unwrapping technique for multiple metamaterial slabs (i.e. the number of met-
amaterial slabs larger than three) can be reasonably chosen according to the frequency at
which the first branch (m = 1 or m = −1) occurs for two metamaterial slabs. According to our
numerous numerical simulations, we find that the increase of the number of metamaterial
slabs, the frequency corresponding to the first branch decreases. Hence the first branch
frequency for multiple metamaterial slabs can be estimated according to the first branch
frequency for few metamaterial slabs. Considering the coupling between different slabs in
multiple metamaterial slabs, the first branch frequency f0 for two metamaterial slabs is first
solved. The starting frequency lower than the first branch frequency for m metamaterial
slabs can be approximately determined as

2f 0
fsm = (14)
m
Noted that as the number of metamaterial slabs increases, the starting frequency tends to
be zero, which is reduced to the conventional procedure.
Waves in Random and Complex Media   5
Downloaded by [Library Services City University London] at 09:17 12 April 2016

Figure 1.  Five slabs of the SRR-Rod structures: (a) geometry; (b) phase; (c) real part of refractive index;
(d) refractive index; (e) permittivity; (f) permeability.
6    Y. Shi et al.
Downloaded by [Library Services City University London] at 09:17 12 April 2016

Figure 1.  (Continued).

3.  Numerical simulation and discussions


In this section, some numerical examples are given to demonstrate good performance of
the proposed retrieval method. Here, S parameters in all numerical examples are solved
using the commercial finite-element solver Ansys HFSS. Each simulation model consisted of
Waves in Random and Complex Media   7
Downloaded by [Library Services City University London] at 09:17 12 April 2016

Figure 2.  Three slabs of the SRR-Rod structures: (a) phase for two slabs of SRR-Rod structures; (b) phase
for three slabs of SRR-Rod structures; (c) refractive index; (d) permittivity; (e) permeability.

a two-port waveguide formed by a pair of both perfect electric conductor (PEC) and perfect
magnetic conductor (PMC) walls. Here, the plane wave with the electric field parallel to z
axis is incident along +y axis.
8    Y. Shi et al.
Downloaded by [Library Services City University London] at 09:17 12 April 2016

Figure 2.  (Continued).

As the first example, consider a structure consisting of five slabs of the split ring resona-
tors (SRRs) and rods, as shown in Figure 1(a). The dimensions of the unit cell element and
material parameters are the same as those in [8]. Figure 1(b) shows the phase ϕ before and
after unwrapping procedure. The real part of the refractive index has been extracted by the
proposed method and compared with the K–K method with three different upper limits of
the integrals, as shown in Figure 1(c). It can be seen that with the increase in the upper limit of
the K–K integral, results solved by the proposed method and the K–K method are in increas-
ing agreement with each other. Because the desirable frequency band is below 20 GHz, the
measurement/simulation of the metamaterial structure should be only conducted in this
frequency band. But the K–K method cannot obtain accurate results using measurement/
simulation S parameters in the concerned frequency band, according to Figure 1(c). By com-
parison, the proposed phase unwrapping method can get accurate result when S parameters
in concerned frequency band are known. Further, the comparisons of the effective refractive
index, permittivity and permeability solved by the proposed method and the K–K method
with the largest upper limit of the integral are given in Figure 1(d)–(f), respectively. The results
obtained by two methods are in good agreement. In addition, the starting frequency of the
proposed phase unwrapping method can be estimated for three slabs of SRR-Rod structures.
The frequency at which the first branch (m = 1) occurs for two slabs of SRR-Rod structures
Waves in Random and Complex Media   9
Downloaded by [Library Services City University London] at 09:17 12 April 2016

Figure 3.  A slab of SRR-CLS structure: (a) geometry; (b) refractive index; (c) permittivity; (d) permeability;
(e) magnitude of S parameters; (f) phase of S parameters.

is f 0 = 9.8GHz, as shown in Figure 2(a). Hence, the starting frequency can be approximately
determined according to (14), i.e. fs3 = 6.53GHz. Figure 2(b) shows ϕ before and after phase
unwrapping procedure with the starting frequency of 6.53 GHz. It can be seen from Figure
2(b) that the branch index m is equal to zero at the starting frequency. Figure 2(c)–(e) shows
10    Y. Shi et al.
Downloaded by [Library Services City University London] at 09:17 12 April 2016

Figure 3.  (Continued).

the effective refractive index, permittivity and permeability, respectively. According to Figure
2(c)–(e), we can observe that two results with the starting frequencies of 0.1 and 6.53 GHz
are in agreement with each other. When the concerned frequency band is far higher than
low frequency, the computational efficiency can be greatly increased with the estimated
starting frequency.
Waves in Random and Complex Media   11
Downloaded by [Library Services City University London] at 09:17 12 April 2016

Figure 4. The sensitivity of the proposed method to phase noise (a) magnitude of S parameters; (b) phase
of S21 with small noise; (c) impedance with small noise; (d) refractive index with small noise; (e) phase of
S21 with large noise; (f) impedance with large noise; (g) refractive index with large noise.

In the following, consider a structure composed of two SRRs and four capacitively loaded
strips (CLSs), which are printed on both sides of the nonmagnetic substrate with relative
permittivity of 2.2 and the thickness of 93 mils, as shown in Figure 3(a). The height along z
12    Y. Shi et al.
Downloaded by [Library Services City University London] at 09:17 12 April 2016

Figure 4. (Continued).

direction, the width along x direction, and the length along y direction of the unit cell are
160, 125, and 290 mils, respectively. All gaps and line widths are 10 mils. The height of the
CLS inclusion is 150 mils, and the length of the full capacity strips is 130 mils. The lengths
of the outer and inner SRRs are 110 and 70 mils, respectively. The real part of refractive
index has been extracted by the proposed method and compared with the K–K method
with three different upper limits of the integrals, as shown in Figure 3(b). It can be seen
that with the increase in upper limit of the K–K integral, results obtained by the proposed
method and the K–K method are in increasing agreement with each other. The effective
refractive index, permittivity and permeability are extracted by the proposed method and
compared with those obtained by K–K method with the largest integral upper limit. As
shown in Figure 3(b)–(d), two results agree well. Figure 3(e) and (f ) show the comparison
of simulated S parameters between the structure and the retrieval effective parameters.
Here, S11 and S21 based on the retrieval effective parameters are calculated by substituting
the extracted z and n into the analytical formulas (1). It can be seen that two results are in
good agreement with each other, which validates the proposed unwrapping algorithm. In
Waves in Random and Complex Media   13
Downloaded by [Library Services City University London] at 09:17 12 April 2016

Figure 4. (Continued).

addition, the sensitivity of the phase unwrapping method to the noise is studied. Figure 4(a)
shows magnitudes of the simulated S parameters. A random noise with a standard uniform
distribution, of which maximum magnitude is 0.05, is introduced to the phase of S21 from
8.14 to 8.72 GHz, where the magnitude of S21 approaches to zero, as shown in Figure 4(b).
Figure 4(c) and (d) show the effective impedance and refractive index, respectively. It can
be seen that the effective impedance and refractive index are insensitive to the small phase
noise. When the magnitude of the random noise is enlarged to 2.0 shown in Figure 4(e), we
can see that the effective impedance and imaginary part of refractive index are still insensi-
tive to the large phase noise, while real part of the refractive index is deteriorated, as shown
in Figure 4(f ) and (g). This is because introduction of the large phase noise to the phase of
S21 leads to incorrect choice of the branch index by the phase unwrapping method. Further,
a random noise with a maximum magnitude of 0.05 is introduced to the magnitude of S21,
as shown in Figure 5(a). According to Figure 5(b) and (c), we can observe that the effective
impedance and real part of refractive index are insensitive to the small magnitude noise,
while imaginary part of the refractive index is sensitive to the small magnitude noise. When
14    Y. Shi et al.
Downloaded by [Library Services City University London] at 09:17 12 April 2016

Figure 5 . The sensitivity of the proposed method to magnitude noise (a) magnitude of S21 with small
noise; (b) impedance with small noise; (c) refractive index with small noise; (d) magnitude of S21 with
large noise; (e) impedance with large noise; (f) refractive index with large noise.
Waves in Random and Complex Media   15
Downloaded by [Library Services City University London] at 09:17 12 April 2016

Figure 5 . (Continued).

the magnitude of the random noise is enlarged to 0.2 shown in Figure 5(d), real parts of the
effective impedance and the refractive index still remain unchanged, while the imaginary
parts of the effective impedance and the refractive index are deteriorated, as shown in
Figure 5(e) and (f ). Considering that the phase unwrapping technique solves the real part
of refractive index, the phase unwrapping technique is insensitive to the magnitude noise.
16    Y. Shi et al.

5. Conclusion
In this paper, we proposed a retrieval method of the effective electromagnetic parameters
for the metamaterial structures. Based on analytical continuation of the effective refractive
index of the metamaterials in the upper half of a complex angular frequency plane, the phase
unwrapping technique has been used to calculate the phase of refractive index, and thus
the branch ambiguity, which occurs in the case of electrically thick metamaterial samples,
has been solved. A method for estimating the starting frequency in the phase unwrapping
method is developed for multiple slabs of the metamaterial structures. Compared with other
retrieval methods which can overcome the branch problem, the retrieval method based on
the proposed phase unwrapping technique is simple and low computational cost. Numerical
Downloaded by [Library Services City University London] at 09:17 12 April 2016

examples demonstrate that the proposed method is accurate and insensitive to magnitude
noise and small phase noise.

Acknowledgment
The authors would like to acknowledge anonymous reviewers for improving this work with their
helpful comments and suggestions.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This work was supported partly by Natural Science Basic Research Plan in Shaanxi Province [grant
number BD11015020008]; Technology Innovation Research Project of the CETC; Fundamental Research
Funds for the Central Universities [grant number SPSZ031410].

References
  [1] Veselago V. The electrodynamics of substances with simultaneously negative values of ε and μ.
Sov. Phys. Usp. 1968;10:509–514.
  [2] Engheta N, Ziolkowski RW. Metamaterials physics and engineering explorations. New Jersey
(NJ): IEEE Press; 2006.
  [3] Cui TJ, Smith DR, Liu RP. Metamaterials theory, design, and applications. New York (NY): Springer;
2010.
  [4] Smith DR. Analytic expressions for the constitutive parameters of magnetoelectric metamaterials.
Phys. Rev. E. 2010;81:036605 (1–17).
  [5] Smith DR, Vier DC, Koschny Th, et al. Electromagnetic parameter retrieval from inhomogeneous
metamaterials. Phys. Rev. E. 2005;71:036617 (1–11).
  [6] Koschny Th, Markoš P, Economou EN, et al. Impact of inherent periodic structure on effective
medium description of left-handed and related metamaterials. Physical Review B. 2005;71:245105
(1–22).
  [7] Chen X, Grzegorczyk TM, Wu BI, et al. Robust method to retrieve the constitutive effective
parameters of metamaterials, Phys. Rev. E. 2004;70:016608 (1–7).
  [8] Zsolt S, Park GH, Ravi H, et al. A unique extraction of metamaterial parameters based on Kramers–
Kronig relationship. IEEE Trans. Microw. Theory and Tech. 2010;58:2646–2653.
Waves in Random and Complex Media   17

  [9] Hansen TV, Mortensen NA, Gregersen AH, et al. A review of the scattering parameter extraction
method with clarification of ambiguity issues in relation to metamaterial homogenization. IEEE
Antennas Propag. Mag. 2013;55:91–106.
  [10] Barroso JJ, Hasar UC. Resolving phase ambiguity in the inverse problem of transmission/reflection
measurement methods. J. Infrared Millimeter Terahertz Waves. 2011;32:857–866.
  [11] Barroso JJ, Hasar UC. Comments on a unique extraction of metamaterial parameters based on
Kramers-Kronig relationship. IEEE Trans. Microw. Theory Tech. 2012;60:1743–1744.
  [12] Barroso JJ, Hasar UC. Constitutive parameters of a metamaterial slab retrieved by the phase
unwrapping method. J. Infrared Millimeter Terahertz Waves. 2012;33:237–244.
  [13] Barroso JJ, de Paula AL. Retrieval of permittivity and permeability of homogeneous materials
from scattering parameters. J. Electromagn. Waves Appl. 2010;24:1563–1574.
  [14] Cao ZX, Yuan FG, Li LH. An automated phase correction algorithm for retrieving permittivity and
permeability of electromagnetic metamaterials. AIP Adv. 2014;4:067115.
Downloaded by [Library Services City University London] at 09:17 12 April 2016

 [15] Luukkonen O, Maslovski SI, Tretyakov SA. A stepwise Nicolson–Ross–Weir-based material


parameter extraction method. IEEE Antennas Wirel. Propag. Lett. 2011;10:1295–1298.
  [16] Hasar UC, Barroso JJ, Sabah C, et al. Resolving phase ambiguity in the inverse problem of
reflection-only measurement methods. Prog. Electromagn. Res. 2012;129:405–420.
  [17] Hasar UC, Barroso JJ, Sabah C, et al. Stepwise technique for accurate and unique retrieval of
electromagnetic properties of bianisotropic metamaterials. J. Opt. Soc. Am. B. 2013;30:1058–
1068.
  [18] Hasar UC, Barroso JJ, Karacali T, et al. Reference-plane-invariant and thickness- and branch-
index-independent retrieval of effective parameters of bi-anisotropic metamaterials. AIP Adv.
2015;5:017123.
  [19] Lucarini V, Saarinen JJ, Peiponen KE, et al. Kramers–Kronig relations in optical materials research.
Berlin: Springer; 2005.

You might also like