What is a metamaterial?

A periodic array of artificial structures (resonators, wires etc.) much smaller than the wavelength, giving electromagnetic properties not found in natural materials. To a wave the fine structure averages out into an apparently uniform medium, described by effective permittivity and permeability. Making both negative gives a negative refractive index.

How is a negative index made?

An array of thin metal wires has negative permittivity below a plasma frequency; an array of split-ring resonators has negative permeability just above its resonance. Where both are negative the index is negative. Here permeability is negative from 10 to 11.95 GHz and permittivity below 15 GHz, so 10–11.95 GHz is the negative-index band.

What should be watched in metamaterial analysis?

The unit cell must be much smaller than the wavelength (about a tenth or less) for the effective-medium view to hold; the 3 mm cell here is about a tenth of the 30 mm wavelength at 10 GHz, borderline. Resonance also brings large loss, so the imaginary part of the index grows near resonance. Real structures are analysed by simulating a unit cell (FDTD or finite elements) and extracting effective parameters from reflection and transmission coefficients.

Metamaterial Electromagnetic Analysis

Category: Electromagnetic Analysis – High Frequency | Revised 2026-10-01

$Metamaterial unit cell electromagnetic simulation showing negative refractive index and split-ring resonator field distribution$

Metamaterial unit cell analysis: Electric field distribution in SRR structure and conceptual diagram of equivalent constitutive parameter extraction

Theory: materials made by averaging fine structure

Overview

🙋

How can a metamaterial have a negative refractive index?

🎓

Refractive index is the square root of the product of permittivity and permeability. Both are positive in ordinary materials. In an array of thin metal wires electrons move freely, so below some frequency they polarise opposite the field and permittivity goes negative. An array of small slit rings (split-ring resonators) magnetises opposite the field just above resonance, making permeability negative. Where both are negative, the physically correct root is negative, giving a negative index. If the structures are much smaller than the wavelength, the wave sees a uniform material described by effective permittivity and permeability.

Effective permeability and permittivity models

$$ \mu_{eff}(\omega) = 1 - \frac{F\,\omega^2}{\omega^2 - \omega_0^2 + i\gamma\omega},\qquad \varepsilon_{eff}(\omega) = 1 - \frac{\omega_p^2}{\omega(\omega + i\gamma_w)},\qquad n = \pm\sqrt{\varepsilon_{eff}\,\mu_{eff}} $$

F is the split-ring filling factor (0.3), f₀ the resonance (10 GHz), γ loss (equivalent to 0.1 GHz) and f_p the wire array's plasma frequency (15 GHz). The sign of the index is chosen so that loss attenuation is positive (Im n > 0).

Negative bands and figure of merit

$$ \mu_{eff} < 0:\ \ \omega_0 < \omega < \frac{\omega_0}{\sqrt{1 - F}},\qquad \varepsilon_{eff} < 0:\ \ \omega < \omega_p,\qquad \mathrm{FOM} = \frac{|\mathrm{Re}\,n|}{\mathrm{Im}\,n} $$

Coffee Break Trivia

Veselago's prediction

In 1968 Veselago predicted theoretically that a material with both permittivity and permeability negative would show strange properties such as reversed refraction. No such natural material existed, and the idea lay dormant. In 1999 Pendry showed split-ring resonators could produce negative permeability, and in 2001 Smith and colleagues combined them with wires to demonstrate negative refraction at microwave frequencies. Research then expanded into areas such as cloaking that steers waves around objects.

Worked examples

Example: effective parameters and index by frequency (3 mm cell)

Frequency	Re ε	Re μ	n	FOM	Cell/λ
9.0 GHz	−1.78	+2.28	−0.02 + 2.01i	0.0	0.090
10.2 GHz	−1.16	−6.26	−2.72 + 0.40i	6.8	0.102
10.5 GHz	−1.04	−2.19	−1.51 + 0.12i	12.7	0.105
11.0 GHz	−0.86	−0.72	−0.79 + 0.05i	14.9	0.110
11.5 GHz	−0.70	−0.23	−0.40 + 0.04i	10.0	0.115
12.5 GHz	−0.44	+0.17	−0.01 + 0.27i	0.0	0.125
16.0 GHz	+0.12	+0.51	+0.25 + 0.00i	61	0.160

At 9 and 12.5 GHz only permittivity is negative, the index is nearly purely imaginary and waves decay without propagating. The negative-permeability band is 10–11.95 GHz (lossless).

🙋

The most negative index is at 10.2 GHz, just above resonance, but the figure of merit is higher at 11 GHz.

🎓

Near resonance permeability becomes strongly negative but loss (the imaginary part) also grows. By magnitude alone 10.2 GHz looks attractive, but the wave decays over a short distance. The figure of merit, the ratio of real to imaginary index, indicates how many wavelengths the wave travels; here it peaks at about 15 near 11 GHz. Practical design balances lowering resonator loss (metal resistance and substrate dielectric loss) against operating somewhat away from resonance. Another caution is that the cell is only about a tenth of a wavelength; larger cells break the effective-medium assumption, and parameters extracted from reflection and transmission then vary with incidence angle and so on.

Analysis workflow

Set resonance and plasma frequencies from the target band.
Simulate a unit cell with periodic boundaries (FDTD or FEM).
Extract effective permittivity and permeability from reflection and transmission (S-parameters).
Check figure of merit and cell-to-wavelength ratio.
Analyse a finite sample to check edge effects and angle dependence.

Coffee Break Trivia

“Extracted index changed with thickness”

In one metamaterial analysis, the effective index extracted from a single unit-cell layer differed greatly from that extracted with three layers. The cell was about a fifth of a wavelength, inter-layer coupling was strong, and the single-layer value didn't represent an effective material. Shrinking the cell to about a tenth of a wavelength made parameters nearly constant regardless of layer count. When extracting effective parameters, confirm they don't change with the number of layers.

Common mistakes

Mistakes and fixes

Mistake	Effect	Fix
Cell too large	Parameters ill-defined	Under λ/10
Wrong square-root branch	Index sign flipped	Choose Im n > 0
Ignoring loss	Performance overestimated	Check FOM
Extracting from one layer only	Coupling missed	Vary layer count
Normal incidence only	Properties change off-normal	Vary incidence angle

🙋

I'd like to learn related topics.

🎓

Tool Name	Developer/Current	Main File Formats
Ansys HFSS	Ansys Inc.	.aedt, .hfss
CST Studio Suite	Dassault Systèmes SIMULIA	.cst
COMSOL Multiphysics	COMSOL AB	.mph