Effective medium approximations

Bruggeman's model

For a mixture of two materials with permittivities ${\displaystyle {\epsilon }_m}$ and ${\displaystyle {\epsilon }_d}$ with corresponding volume fractions ${\displaystyle c_m}$ and ${\displaystyle c_d}$, D.A.G. Bruggeman proposed a formula in 1935 of the following form:

Here the positive sign before the square root must be altered to a negative sign in some cases in order to get the correct imaginary part of effective complex permittivity which is related with electromagnetic wave attenuation. The formula is symmetric with respect to swapping the 'd' and 'm' roles. This formula is based on the equality

where ${\displaystyle \mathrm{\Delta }\mathrm{\Phi }}$ is the jump of electric displacement flux all over the integration surface, ${\displaystyle E_n(\mathbf{r})}$ is the component of microscopic electric field normal to the integration surface, ${\displaystyle {\epsilon }_r(\mathbf{r})}$ is the local relative complex permittivity which takes the value ${\displaystyle {\epsilon }_m}$ inside the picked metal particle, the value ${\displaystyle {\epsilon }_d}$ inside the picked dielectric particle and the value ${\displaystyle {\epsilon }_{\mathrm{e}\mathrm{f}\mathrm{f}}}$ outside the picked particle, ${\displaystyle E_0}$ is the normal component of the macroscopic electric field. Formula (4) comes out of Maxwell's equality ${\displaystyle \mathrm{div}({\epsilon }_r\mathbf{E})=0}$. Thus only one picked particle is considered in Bruggeman's approach. The interaction with all the other particles is taken into account only in a mean field approximation described by ${\displaystyle {\epsilon }_{\mathrm{e}\mathrm{f}\mathrm{f}}}$. Formula (3) gives a reasonable resonant curve for plasmon excitations in metal nanoparticles if their size is 10 nm or smaller. However, it is unable to describe the size dependence for the resonant frequency of plasmon excitations that are observed in experiments

Formulas

Without any loss of generality, we shall consider the study of the effective conductivity (which can be either dc or ac) for a system made up of spherical multicomponent inclusions with different arbitrary conductivities. Then the Bruggeman formula takes the form:

Circular and spherical inclusions

In a system of Euclidean spatial dimension ${\displaystyle n}$ that has an arbitrary number of components, the sum is made over all the constituents. ${\displaystyle {\delta }_i}$ and ${\displaystyle {\sigma }_i}$ are respectively the fraction and the conductivity of each component, and ${\displaystyle {\sigma }_e}$ is the effective conductivity of the medium. (The sum over the ${\displaystyle {\delta }_i}$'s is unity.)

Elliptical and ellipsoidal inclusions

This is a generalization of Eq. (1) to a biphasic system with ellipsoidal inclusions of conductivity ${\displaystyle \sigma }$ into a matrix of conductivity ${\displaystyle {\sigma }_m}$. The fraction of inclusions is ${\displaystyle \delta }$ and the system is ${\displaystyle n}$ dimensional. For randomly oriented inclusions,

where the ${\displaystyle L_j}$'s denote the appropriate doublet/triplet of depolarization factors which is governed by the ratios between the axis of the ellipse/ellipsoid. For example: in the case of a circle (${\displaystyle L_1=1/2}$, ${\displaystyle L_2=1/2}$) and in the case of a sphere (${\displaystyle L_1=1/3}$, ${\displaystyle L_2=1/3}$, ${\displaystyle L_3=1/3}$). (The sum over the ${\displaystyle L_j}$ 's is unity.)

The most general case to which the Bruggeman approach has been applied involves bianisotropic ellipsoidal inclusions.

Derivation

The figure illustrates a two-component medium. Consider the cross-hatched volume of conductivity ${\displaystyle {\sigma }_1}$, take it as a sphere of volume ${\displaystyle V}$ and assume it is embedded in a uniform medium with an effective conductivity ${\displaystyle {\sigma }_e}$. If the electric field far from the inclusion is ${\displaystyle \overline{E_0}}$ then elementary considerations lead to a dipole moment associated with the volume

This polarization produces a deviation from ${\displaystyle \overline{E_0}}$. If the average deviation is to vanish, the total polarization summed over the two types of inclusion must vanish. Thus

where ${\displaystyle {\delta }_1}$ and ${\displaystyle {\delta }_2}$ are respectively the volume fraction of material 1 and 2. This can be easily extended to a system of dimension ${\displaystyle n}$ that has an arbitrary number of components. All cases can be combined to yield Eq. (1).

Eq. (1) can also be obtained by requiring the deviation in current to vanish. It has been derived here from the assumption that the inclusions are spherical and it can be modified for shapes with other depolarization factors; leading to Eq. (2).

A more general derivation applicable to bianisotropic materials is also available.

Modeling of percolating systems

The main approximation is that all the domains are located in an equivalent mean field. Unfortunately, it is not the case close to the percolation threshold where the system is governed by the largest cluster of conductors, which is a fractal, and long-range correlations that are totally absent from Bruggeman's simple formula. The threshold values are in general not correctly predicted. It is 33% in the EMA, in three dimensions, far from the 16% expected from percolation theory and observed in experiments. However, in two dimensions, the EMA gives a threshold of 50% and has been proven to model percolation relatively well.

Maxwell Garnett equation

In the Maxwell Garnett approximation, the effective medium consists of a matrix medium with ${\displaystyle {\epsilon }_m}$ and inclusions with ${\displaystyle {\epsilon }_i}$. Maxwell Garnett was the son of physicist William Garnett, and was named after Garnett's friend, James Clerk Maxwell. He proposed his formula to explain coloured pictures that are observed in glasses doped with metal nanoparticles. His formula has a form

where ${\displaystyle {\epsilon }_{\text{eff}}}$ is effective relative complex permittivity of the mixture, ${\displaystyle {\epsilon }_d}$ is relative complex permittivity of the background medium containing small spherical inclusions of relative permittivity ${\displaystyle {\epsilon }_m}$ with volume fraction of ${\displaystyle c_m\ll 1}$. This formula is based on the equality

where ${\displaystyle {\epsilon }_0}$ is the absolute permittivity of free space and ${\displaystyle p_m}$ is electric dipole moment of a single inclusion induced by the external electric field E. However this equality is good only for homogeneous medium and ${\displaystyle {\epsilon }_d=1}$. Moreover, the formula (1) ignores the interaction between single inclusions. Because of these circumstances, formula (1) gives too narrow and too high resonant curve for plasmon excitations in metal nanoparticles of the mixture.

Formula

The Maxwell Garnett equation reads:

where ${\displaystyle {\epsilon }_{\mathrm{e}\mathrm{f}\mathrm{f}}}$ is the effective dielectric constant of the medium, ${\displaystyle {\epsilon }_i}$ of the inclusions, and ${\displaystyle {\epsilon }_m}$ of the matrix; ${\displaystyle {\delta }_i}$ is the volume fraction of the inclusions.

The Maxwell Garnett equation is solved by:

so long as the denominator does not vanish. A simple MATLAB calculator using this formula is as follows.

<span id="mwAQE"></span><span class="c" id="mwAQI">% This simple MATLAB calculator computes the effective dielectric</span> <span class="c" id="mwAQM">% constant of a mixture of an inclusion material in a base medium</span> <span class="c" id="mwAQQ">% according to the Maxwell-Garnett theory.</span> <span class="c" id="mwAQU">% INPUTS:</span> <span class="c" id="mwAQY">% eps_base: dielectric constant of base material;</span> <span class="c" id="mwAQc">% eps_incl: dielectric constant of inclusion material;</span> <span class="c" id="mwAQg">% vol_incl: volume portion of inclusion material;</span> <span class="c" id="mwAQk">% OUTPUT:</span> <span class="c" id="mwAQo">% eps_mean: effective dielectric constant of the mixture.</span> <span class="k" id="mwAQs">function</span><span class="w" id="mwAQw"> </span>eps_mean<span class="w" id="mwAQ0"> </span><span class="p" id="mwAQ4">=</span><span class="w" id="mwAQ8"> </span><span class="nf" id="mwARA">MaxwellGarnettFormula</span><span class="p" id="mwARE">(</span>eps_base, eps_incl, vol_incl<span class="p" id="mwARI">)</span> <span class="w" id="mwARM"> </span><span class="n" id="mwARQ">small_number_cutoff</span><span class="w" id="mwARU"> </span><span class="p" id="mwARY">=</span><span class="w" id="mwARc"> </span><span class="mf" id="mwARg">1e-6</span><span class="p" id="mwARk">;</span> <span class="w" id="mwARo"> </span><span class="k" id="mwARs">if</span><span class="w" id="mwARw"> </span><span class="n" id="mwAR0">vol_incl</span><span class="w" id="mwAR4"> </span><span class="o" id="mwAR8"><</span><span class="w" id="mwASA"> </span><span class="mi" id="mwASE">0</span><span class="w" id="mwASI"> </span><span class="o" id="mwASM">||</span><span class="w" id="mwASQ"> </span><span class="n" id="mwASU">vol_incl</span><span class="w" id="mwASY"> </span><span class="o" id="mwASc">></span><span class="w" id="mwASg"> </span><span class="mi" id="mwASk">1</span> <span class="w" id="mwASo"> </span><span class="nb" id="mwASs">disp</span><span class="p" id="mwASw">(</span><span class="s" id="mwAS0">'WARNING: volume portion of inclusion material is out of range!'</span><span class="p" id="mwAS4">);</span> <span class="w" id="mwAS8"> </span><span class="k" id="mwATA">end</span> <span class="w" id="mwATE"> </span><span class="n" id="mwATI">factor_numer</span><span class="w" id="mwATM"> </span><span class="p" id="mwATQ">=</span><span class="w" id="mwATU"> </span><span class="mi" id="mwATY">2</span><span class="w" id="mwATc"> </span><span class="o" id="mwATg">*</span><span class="w" id="mwATk"> </span><span class="p" id="mwATo">(</span><span class="mi" id="mwATs">1</span><span class="w" id="mwATw"> </span><span class="o" id="mwAT0">-</span><span class="w" id="mwAT4"> </span><span class="n" id="mwAT8">vol_incl</span><span class="p" id="mwAUA">)</span><span class="w" id="mwAUE"> </span><span class="o" id="mwAUI">*</span><span class="w" id="mwAUM"> </span><span class="n" id="mwAUQ">eps_base</span><span class="w" id="mwAUU"> </span><span class="o" id="mwAUY">+</span><span class="w" id="mwAUc"> </span><span class="p" id="mwAUg">(</span><span class="mi" id="mwAUk">1</span><span class="w" id="mwAUo"> </span><span class="o" id="mwAUs">+</span><span class="w" id="mwAUw"> </span><span class="mi" id="mwAU0">2</span><span class="w" id="mwAU4"> </span><span class="o" id="mwAU8">*</span><span class="w" id="mwAVA"> </span><span class="n" id="mwAVE">vol_incl</span><span class="p" id="mwAVI">)</span><span class="w" id="mwAVM"> </span><span class="o" id="mwAVQ">*</span><span class="w" id="mwAVU"> </span><span class="n" id="mwAVY">eps_incl</span><span class="p" id="mwAVc">;</span> <span class="w" id="mwAVg"> </span><span class="n" id="mwAVk">factor_denom</span><span class="w" id="mwAVo"> </span><span class="p" id="mwAVs">=</span><span class="w" id="mwAVw"> </span><span class="p" id="mwAV0">(</span><span class="mi" id="mwAV4">2</span><span class="w" id="mwAV8"> </span><span class="o" id="mwAWA">+</span><span class="w" id="mwAWE"> </span><span class="n" id="mwAWI">vol_incl</span><span class="p" id="mwAWM">)</span><span class="w" id="mwAWQ"> </span><span class="o" id="mwAWU">*</span><span class="w" id="mwAWY"> </span><span class="n" id="mwAWc">eps_base</span><span class="w" id="mwAWg"> </span><span class="o" id="mwAWk">+</span><span class="w" id="mwAWo"> </span><span class="p" id="mwAWs">(</span><span class="mi" id="mwAWw">1</span><span class="w" id="mwAW0"> </span><span class="o" id="mwAW4">-</span><span class="w" id="mwAW8"> </span><span class="n" id="mwAXA">vol_incl</span><span class="p" id="mwAXE">)</span><span class="w" id="mwAXI"> </span><span class="o" id="mwAXM">*</span><span class="w" id="mwAXQ"> </span><span class="n" id="mwAXU">eps_incl</span><span class="p" id="mwAXY">;</span> <span class="w" id="mwAXc"> </span><span class="k" id="mwAXg">if</span><span class="w" id="mwAXk"> </span><span class="nb" id="mwAXo">abs</span><span class="p" id="mwAXs">(</span><span class="n" id="mwAXw">factor_denom</span><span class="p" id="mwAX0">)</span><span class="w" id="mwAX4"> </span><span class="o" id="mwAX8"><</span><span class="w" id="mwAYA"> </span><span class="n" id="mwAYE">small_number_cutoff</span> <span class="w" id="mwAYI"> </span><span class="nb" id="mwAYM">disp</span><span class="p" id="mwAYQ">(</span><span class="s" id="mwAYU">'WARNING: the effective medium is singular!'</span><span class="p" id="mwAYY">);</span> <span class="w" id="mwAYc"> </span><span class="n" id="mwAYg">eps_mean</span><span class="w" id="mwAYk"> </span><span class="p" id="mwAYo">=</span><span class="w" id="mwAYs"> </span><span class="mi" id="mwAYw">0</span><span class="p" id="mwAY0">;</span> <span class="w" id="mwAY4"> </span><span class="k" id="mwAY8">else</span> <span class="w" id="mwAZA"> </span><span class="n" id="mwAZE">eps_mean</span><span class="w" id="mwAZI"> </span><span class="p" id="mwAZM">=</span><span class="w" id="mwAZQ"> </span><span class="n" id="mwAZU">eps_base</span><span class="w" id="mwAZY"> </span><span class="o" id="mwAZc">*</span><span class="w" id="mwAZg"> </span><span class="n" id="mwAZk">factor_numer</span><span class="w" id="mwAZo"> </span><span class="o" id="mwAZs">/</span><span class="w" id="mwAZw"> </span><span class="n" id="mwAZ0">factor_denom</span><span class="p" id="mwAZ4">;</span> <span class="w" id="mwAZ8"> </span><span class="k" id="mwAaA">end</span> <span class="k" id="mwAaE">end</span>

Derivation

For the derivation of the Maxwell Garnett equation we start with an array of polarizable particles. By using the Lorentz local field concept, we obtain the Clausius-Mossotti relation: $${\displaystyle \frac{\epsilon -1}{\epsilon +2}=\frac{4\pi }{3}\sum _j{N}_j{\alpha }_j}$$ Where ${\displaystyle N_j}$ is the number of particles per unit volume. By using elementary electrostatics, we get for a spherical inclusion with dielectric constant ${\displaystyle {\epsilon }_i}$ and a radius ${\displaystyle a}$ a polarisability ${\displaystyle \alpha }$: $${\displaystyle \alpha =\left(\frac{{\epsilon }_i-1}{{\epsilon }_i+2}\right)a^3}$$ If we combine ${\displaystyle \alpha }$ with the Clausius Mosotti equation, we get: $${\displaystyle \left(\frac{{\epsilon }_{\mathrm{e}\mathrm{f}\mathrm{f}}-1}{{\epsilon }_{\mathrm{e}\mathrm{f}\mathrm{f}}+2}\right)={\delta }_i\left(\frac{{\epsilon }_i-1}{{\epsilon }_i+2}\right)}$$ Where ${\displaystyle {\epsilon }_{\mathrm{e}\mathrm{f}\mathrm{f}}}$ is the effective dielectric constant of the medium, ${\displaystyle {\epsilon }_i}$ of the inclusions; ${\displaystyle {\delta }_i}$ is the volume fraction of the inclusions.
As the model of Maxwell Garnett is a composition of a matrix medium with inclusions we enhance the equation:

Validity

In general terms, the Maxwell Garnett EMA is expected to be valid at low volume fractions ${\displaystyle {\delta }_i}$, since it is assumed that the domains are spatially separated and electrostatic interaction between the chosen inclusions and all other neighbouring inclusions is neglected. The Maxwell Garnett formula, in contrast to Bruggeman formula, ceases to be correct when the inclusions become resonant. In the case of plasmon resonance, the Maxwell Garnett formula is correct only at volume fraction of the inclusions ${\displaystyle {\delta }_i<{10}^{-5}}$. The applicability of effective medium approximation for dielectric multilayers and metal-dielectric multilayers have been studied, showing that there are certain cases where the effective medium approximation does not hold and one needs to be cautious in application of the theory.

Generalization of the Maxwell Garnett equation to describe the nanoparticle size distribution

Maxwell Garnett equation describes optical properties of nanocomposites which consist in a collection of perfectly spherical nanoparticles. All these nanoparticles must have the same size. However, due to confinement effect, the optical properties can be influenced by the nanoparticles size distribution. As shown by Battie et al., the Maxwell Garnett equation can be generalized to take into account this distribution.

${\displaystyle \frac{({\epsilon }_{\text{eff}}-{\epsilon }_m)}{{\epsilon }_{\text{eff}}-2{\epsilon }_m}=\frac{3i{\lambda }^3}{16{\pi }^2{\epsilon }_m^{1.5}}\frac{f}{R_m^3}\int P(R)a_1(R)dR}$

${\displaystyle R}$ and ${\displaystyle P(R)}$ are the nanoparticle radius and size distribution, respectively. ${\displaystyle R_m}$ and ${\displaystyle f}$ are the mean radius and the volume fraction of the nanoparticles, respectively. ${\displaystyle a_1}$ is the first electric Mie coefficient. This equation reveals that the classical Maxwell Garnett equation gives a false estimation of the volume fraction nanoparticles when the size distribution cannot be neglected.

Generalization to include shape distribution of nanoparticles

The Maxwell Garnett equation only describes the optical properties of a collection of perfectly spherical nanoparticles. However, the optical properties of nanocomposites are sensitive to the nanoparticles shape distribution. To overcome this limit, Y. Battie et al. have developed the shape distributed effective medium theory (SDEMT). This effective medium theory enables to calculate the effective dielectric function of a nanocomposite which consists in a collection of ellipsoïdal nanoparticles distributed in shape.

${\displaystyle {\epsilon }_{\text{eff}}=\frac{(1-f){\epsilon }_m+f\beta {\epsilon }_i}{1-f+f\beta }}$

with ${\displaystyle \beta =\frac{1}{3}\iint P(L_1,L_2)\sum _{i=1}^3\frac{{\epsilon }_m}{{\epsilon }_m+L_i({\epsilon }_i-{\epsilon }_m)}d{L}_{1}d{L}_2}$

The depolarization factors (${\displaystyle L_1,L_2,L_3}$) only depend on the shape of nanoparticles. ${\displaystyle P(L_1,L_2)}$ is the distribution of depolarization factors. ${\displaystyle f}$ is the volume fraction of the nanoparticles.

The SDEMT theory was used to extract the shape distribution of nanoparticles from absorption or ellipsometric spectra.

Formula describing size effect

A new formula describing size effect was proposed. This formula has a form $${\displaystyle {\epsilon }_{\text{eff}}=\frac{1}{4}\left(H_{\epsilon }+i\sqrt{-H_{\epsilon }^2-8{\epsilon }_m{\epsilon }_{d}J(k_{m}a)}\right),}$$

$${\displaystyle J(x)=2\frac{1-x\cot (x)}{x^2+x\cot (x)-1},}$$ where a is the nanoparticle radius and ${\displaystyle k_m=\sqrt{{\epsilon }_m{\mu }_m}\omega /c}$ is wave number. It is supposed here that the time dependence of the electromagnetic field is given by the factor ${\displaystyle \mathrm{e}\mathrm{x}\mathrm{p}(-i\omega t).}$ In this paper Bruggeman's approach was used, but electromagnetic field for electric-dipole oscillation mode inside the picked particle was computed without applying quasi-static approximation. Thus the function ${\displaystyle J(k_{m}a)}$ is due to the field nonuniformity inside the picked particle. In quasi-static region (${\displaystyle k_{m}a\ll 1}$, i.e. ${\displaystyle a\le 10\mathrm{n}\mathrm{m}}$ for Ag${\displaystyle )}$ this function becomes constant ${\displaystyle J(k_{m}a)=1}$ and formula (5) becomes identical with Bruggeman's formula.

Effective permeability formula

Formula for effective permeability of mixtures has a form

$${\displaystyle H_{\mu }=(2-3c_m){\mu }_d-(1-3c_m){\mu }_{m}J(k_{m}a).}$$

Here ${\displaystyle {\mu }_{\text{eff}}}$ is effective relative complex permeability of the mixture, ${\displaystyle {\mu }_d}$ is relative complex permeability of the background medium containing small spherical inclusions of relative permeability ${\displaystyle {\mu }_m}$ with volume fraction of ${\displaystyle c_m\ll 1}$. This formula was derived in dipole approximation. Magnetic octupole mode and all other magnetic oscillation modes of odd orders were neglected here. When ${\displaystyle {\mu }_m={\mu }_d}$ and ${\displaystyle k_{m}a\ll 1}$ this formula has a simple form

Effective medium theory for resistor networks

For a network consisting of a high density of random resistors, an exact solution for each individual element may be impractical or impossible. In such case, a random resistor network can be considered as a two-dimensional graph and the effective resistance can be modelled in terms of graph measures and geometrical properties of networks. Assuming, edge length is much less than electrode spacing and edges to be uniformly distributed, the potential can be considered to drop uniformly from one electrode to another. Sheet resistance of such a random network (${\displaystyle R_{sn}}$) can be written in terms of edge (wire) density (${\displaystyle N_E}$), resistivity (${\displaystyle \rho }$), width (${\displaystyle w}$) and thickness (${\displaystyle t}$) of edges (wires) as: