You are on page 1of 12

1 The Relationship Between Loss, Conductivity, and Dielectric Constant General Expressions The question has been asked

how loss, conductivity, and dielectric constant are interrelated. Answering this question requires a fairly extensive review of basic electromagnetics. First, assume that one has a piece of arbitrary material. This material is made of atoms, molecules, or ions. Within this material exist electrons, either bound to individual atoms or free to move about. An electric field is applied across the object. The electrons will naturally want to move because of the electric field. The conduction electric current density (a measure of the flow of electrons) varies directly with the strength of the electric field. Thus
Jc = σ s E

(1)

where σ s is a constant of proportionality, and it is called “conductivity.” The conductivity provides a measure of how fast an electron can flow through a material. It is defined as

σ s = − qµ e

(2)

where q is the charge and µe is the electric mobility (not the permeability) of the medium. Likewise, the electric flux density varies linearly with the application of the electric field so that
D =εE

(3)

Here, ε is the constant of proportionality, and it is called “permittivity.” The time-harmonic version of Maxwell’s equations states that

∇ × H = J + jω D

(4)

J is the electric current density, and it has two parts. The first part is the impressed electric current density, J i (that is, J i is an excitation to the system by an outside source), and the second part is the aforementioned conduction electric current density, J c , caused by the application of an external electric field. Thus, we have ∇ × H = J i + J c + jω D ∇ × H = J i + σ s E + jω D

(5) (6)

In most materials there exists at least one of three types of electric dipoles. Any kind of dipole exhibits a polarity; that is, one side of the dipole can be described as being Chris Bishop 11/13/2001

The degree to which the dipole is out of phase with the incident electric field and the losses that ensue determine how large the imaginary part of the permittivity is as a function of material and frequency. and the other side can be described as being positively charged. and the less Chris Bishop 11/13/2001 . water has a net electric polarity. Hence. the dipole rotates to align itself with the field. The result is that the electron cloud shifts its position and is no longer centered about the nucleus. The larger the imaginary part. The three types of dipoles are as follows. the dipoles align with the field. 2. As time passes. ε r is in general a complex quantity. For instance. and a positive oxygen atom is on the other side. Most atoms have a cloud of electrons surrounding the nucleus. This relationship can be mathematically described as D = ε0 E + ε0χe E (7) The term χ e is known as the electric susceptibility and serves as a proportionality constant between the electric field and the portion of the electric flux density caused by the presence of the dielectric. To understand why. NaCl. When an external electric field is applied. and the dipole must rotate again to remain aligned with the correct polarity. 3. consider an alternating electric field applied to a dipole. This action causes a term to be added to the electric flux density that has the same vector direction as the applied field. water is bound in such a way that the two negative hydrogen atoms are on one side of the molecule. Ions have inherently oppositely charged parts. As it rotates. Molecules arranged in such a way as to exhibit an imbalance of charge. For instance. the application of an electric field causes the electrons to react and move much more quickly than the nucleus can react. Since the mass of an electron is much less than the mass of the nucleus. has a positive sodium atom (Na+) and a negative chlorine atom (Cl-). When the field first strikes the dipole. energy is lost through the generation of heat (friction) as well as the acceleration and deceleration of the rotational motion of the dipole. Hence. One can rewrite the equation as D = ε 0 (1 + χ e )E (8) or D = ε 0ε r E (9) where ε r is known as the relative permittivity of the medium. 1. the atom ends with the positively charged nucleus on one side and the negatively charged electron cloud on the other side. the electric field reverses its direction. the more energy is being dissipated through motion. table salt.2 negatively charged.

we now have that ∇ × H = J i + σ s E + jω (ε ′ − jε ′′)E ∇ × H = J i + (σ s + ωε ′′)E + jωε ′ E ∇ × H = J i + σ e E + jω ε ′ E (10) (11) (12) (13) In this last step. although it might be labeled as merely “conductivity. we have now (15) σ   ∇ × H = J i + jωε ′1 − j e  E ωε ′   ∇ × H = J i + jωε ′(1 − j tan δ e )E (16) (17) Here. Thus σe = σs +σa Again returning to Maxwell’s equation (13). and we can define the last term to be conductivity due to an alternating field. the following convention is used. the imaginary part of the relative permittivity directly relates to loss in the system.” The first term on the right-hand side of the above equation is the static conductivity. ε 0 ε r = ε ′ − jε ′′ Returning to Maxwell’s equation (6). we have defined the loss tangent. tan δ e as tan δ e = σe ωε ′ (18) We can also expand Maxwell’s equation (16) as σ ε ′′   ∇ × H = J i + jωε ′1 − j s − j  E ωε ′ ε′   Chris Bishop 11/13/2001 (19) .3 energy is available to propagate past the dipole. we have defined an effective conductivity. To represent the real and imaginary parts of the absolute permittivity. Thus. σ e = σ s + ωε ′′ (14) The effective conductivity is the value that is usually specified in data sheets.

Maxwell’s equation (19) becomes ε ′′   ∇ × H ≈ J i + jωε ′1 − j  E ε′   (23) We turn now to calculating the power absorbed and transmitted by a medium. . in conductors.4 The first term. describes loss due to collisions of electrons with other electrons and ωε ′ atoms. At first glance it seems strange that a term that approaches infinity in the numerator describes a low loss structure. In dielectrics. the real part of the permittivity is usually equal to the permittivity of free space. dominates the other term of (19). Jc = σ s E This last equation highlights the fact that two terms contribute to the loss tangent. s . (20) view the electric field as a function of the current density. Maxwell’s equation (19) reduces to ∇ × H ≈ J i + ( jωε 0 + σ s )E (22) For a dielectric. Semiconductors maintain a relative balance between the two terms. σs . That is. E= Jc (21) σs Now infinite conductivity makes sense. For instance. the effective conductivity is due almost entirely to the collisions of electrons. then charges flow very easily without many collisions. and the imaginary part is usually zero. We can begin with Maxwell’s equations and find the following relationships. instead of viewing the current density as a function of the electric field. Thus. and the polarization dependent term is dropped. In metals. and the first term is dropped from the calculation. when an effective conductivity is specified on a data sheet. As one might expect. ωε ′ ε′ ε ′′ The term of (19) describes how much energy supplied by an external electric ε′ field is dissipated as motion and heat. this term of σ ε ′′ (19). These Chris Bishop 11/13/2001 . but it must be remembered that infinite conductivity implies zero electric field (and finite current density). the effective conductivity is due almost entirely to polarization loss (dipole motion). For a metal. if the static conductivity is high (copper has σ s = 5. it is useful to remember that it arises from two sources.8 x10 7 S / m ). this term usually dominates the first term.

Here we ignore the possibility of complex permeability and rewrite the equation set as follows. and Wm and We represent stored energies. That is. one can not merely replace by j ω in the result ∂t because of the non-linear nature of the equations (products of fields). Note that if these equations were derived using time derivatives. For this reason. Ps = Pe + Pd + j 2ω (Wm + We ) Ps = − * * 1 ∫∫∫ H ⋅ M i + E ⋅ J i dv 2 V (30) ( ) (31) Chris Bishop 11/13/2001 . but if µ or ε are complex. usually represented as P = E × H . the ∂ results would be different. The real parts of these terms can be extracted and added to the dissipated power. Pe represents the complex exiting (transmitted) power. then either or both terms may be complex. then the results must be averaged over time to obtain the results here.5 equations are derived assuming that phasors represent the fields. If one uses the usual (non-phasor) derivation. Pd represents the real dissipated power. Usually these last two terms are strictly imaginary. * Ps = Pe + Pd + j 2ω (Wm + We ) Ps = − * * 1 H ⋅ M i + E ⋅ J i dv 2 ∫∫∫ V (24) ( ) (25) Pe = * 1 ∫∫ E × H ⋅ d s 2 S ( ) (26) 2 Pd = 1 1 1 1 ∫∫∫ J ⋅ E dv = 2 ∫∫∫ σ s E ⋅ E dv = 2 ∫∫∫ σ s E dv = 2 ∫∫∫ σ s dv 2 V V V V (27) * * 2 J Wm = 1 µH 4 ∫∫∫ V 2 dv (28) We = 2 1 ∫∫∫ ε E dv 4 V (29) In these equations. and a dependence of e jω t is suppressed. the Poynting vector. Ps represents the complex supplied power. here has the form P = E × H .

we choose a block of dielectric material that has a plane wave propagating inside. it would seem that discussion about a “source-free” medium could not apply. are zero. Thus. we note that non-zero but finite static conductivity implies that charges are present within the medium. it is puzzling to see that the dissipated power varies directly with the static conductivity. the problem can be treated by the same techniques used to treat source-free media. However. This fact is true because non-zero finite static conductivity implies that charges take a finite time to travel through the medium. As mentioned before. the time-averaged charge is zero.” By “source-free” these books mean. Pd = 1 2 ∫∫∫ V  σ s + ωε ′′     σ2  J s   2 dv (36) We now wish to compare the dissipated power with the exiting power and the stored powers. This wave originated from outside the block. The plane wave enters at z = 0 and exits at z = z 0 . because the charges can be expressed in terms of the electric field ( J c = σ s E ). That is. This fact is very Chris Bishop 11/13/2001 . the time-averaged charge is again zero. Infinite static conductivity implies that charges progress instantaneously through the medium. When considering a material with a finite non-zero conductivity. so that the time-averaged charge is non-zero. By “lossy” these same books mean that there is polarization loss present in the material.6 Pe = * 1 ∫∫ E × H ⋅ d s 2 S 2 2 1 (σ s + ωε ′′) E dv = 1 ∫∫∫ σ e E dv 2 ∫∫∫ 2 V V ( ) (32) Pd = (33) Wm = 1 µH 4 ∫∫∫ V 2 dv (34) We = 2 1 ε ′ E dv 4 ∫∫∫ V (35) We are concerned here primarily with the dissipated power term. and Wm and We are the stored energies. Pd . in part. Rewriting Pd in terms of the current density helps the intuition. J i and M i . As an example. Pd is the amount of power dissipated in the medium. thus. that the static conductivity is either zero or infinite. Before we proceed. That is. Pe represents both the entering and the exiting power depending on the surface at which it is evaluated. Often books talk about a “source-free lossy medium. Zero static conductivity implies that free charges never enter the medium. so the impressed sources. the permittivity has a non-zero imaginary part.

7 fortunate. As it is. Chris Bishop 11/13/2001 . otherwise materials with non-zero static conductivities could only be treated numerically. The wave equation that we must solve is ∇ 2 E + k 2 E = jω µ σ s E (37) or ∇ 2 E + k 2 − jω µ σ s E = 0 ( ) (38) Recalling that k 2 = ω 2 µ ε = ω 2 µ (ε ′ − j ε ′′) one obtains  ∇ 2 E + ω 2 µ ε ′ 1 −   or  σ  ∇ 2 E + ω 2 µ ε ′ 1 − j e  E = 0  ωε′   (39)  ε ′′ σ   j + s   E = 0  ε ′ ω ε ′    (40) (41) or ∇ 2 E + ω 2 µ ε ′ (1 − j tan δ e )E = 0 (42) or ∇2 E + γ 2 E = 0 where (43) γ 2 = ω 2 µ ε ′ (1 − j tan δ e ) A solution to the wave equation is ˆ E = E0 e − j γ z x (44) (45) γ can be found as follows. we will now derive the closed-form solution to a plane wave traveling in a medium with finite non-zero static conductivity and with polarization loss.

8 γ = ω 2 µ ε ′ (1 − j tan δ e ) (46) γ = ω µ ε ′ (1 + tan 2 δ e ) 4  cos 1   δe 2 − j sin δe   2 (47) γ =ω µε′ δ δ  1   cos e − j sin e  2 2 cos δ e   1 + cos δ e 1 − cos δ e −j 2 cos δ e  2 cos δ e     (48) γ = ω µ ε ′  (49) The solution to the wave equation is then E = E0 e −ω µ ε ′ 1−cos δ e z 2 cos δ e e − j ω µ ε′ 1+ cos δ e z 2 cos δ e ˆ x (50) Letting α =ω µε′ and 1 − cos δ e 2 cos δ e (51) β =ω µε′ then 1 + cos δ e 2 cos δ e (52) γ = β − jα ˆ E = E0 e −α z e − j β z x (53) (54) Notice that if σ s = 0 and ε ′′ = 0 then δ e = 0 so that α =0 (55) (56) β =k and Chris Bishop 11/13/2001 .

a condition that is not generally true. The magnetic field can be found from ∇ × E = − jω µ H (58) or ˆ − j ω µ H = − j γ E0 e − j γ z y (59) or H= γ ˆ E0 e − j γ z y ωµ (60) This expression.9 ˆ E = E0 e − j k z x (57) which is the usual expression for a plane wave in lossless space. Pe = * 1 1 2 β + j α −2α z ˆ ∫∫ E × H ⋅ d s = 2 ∫∫ E0 ω µ e z ⋅ d s 2 S S 2 1 1 2 − 2α z ∫∫∫ σ e E dv = 2 ∫∫∫ σ e E0 e dv 2 V V 2 1 1 ε′ 2 − 2α z ∫∫∫ µ H dv = 4 ∫∫∫ cos δ e E0 e dv 4 V V ( ) (64) Pd = (65) Wm = (66) Chris Bishop 11/13/2001 . (60). is only equal to H= 1 ε ˆ ˆ E0 e − j γ z y = E0 e − j γ z y η µ (61) when σ s = 0 . ˆ E = E0 e −α z e − j β z x (62) (63) H= β − jα ˆ E0 e −α z e − j β z y ωµ The next step is to compute the desired powers. We now summarize the equations of the fields.

The power entering the material at z = 0 : Pin = − Pe =− z =0 1 A 2 β + jα 2 β + j α − 2α z ˆ ∫∫ E0 ω µ e z ⋅ d s = 2 E0 ω µ 2 S (68) The power exiting the material at z = z 0 : Pout = Pe = z = z0 1 A 2 β + j α −2α z 2 β + j α − 2α z ˆ ∫∫ E0 ω µ e z ⋅ d s = 2 E0 ω µ e 0 2 S (69) The power dissipated inside the material as heat: Pd = 1 A − 2α z 2 − 2α z 2 ∫∫∫ σ e E0 e dv = 4α σ e E0 1 − e 0 2 V ( ) (70) The power stored in magnetic fields: Wm = 1 A ε′ ε′ − 2α z 2 − 2α z 2 ∫∫∫ cos δ e E0 e dv = 8α cos δ e E0 1 − e 0 4 V ( ) (71) The power stored in electric fields: We = 1 A − 2α z 2 − 2α z 2 ∫∫∫ ε ′ E0 e dv = 8α ε ′ E0 1 − e 0 4 V ( ) (72) Note that Pin and Pout appear on the same side of equation (30).10 We = 2 1 1 2 − 2α z ∫∫∫ ε ′ E dv = 4 ∫∫∫ ε ′ E0 e dv 4 V V (67) For a block of material having length. and cross-sectional area. and stored powers equal to the input power. the integrals evaluate as follows. dissipated. Taking the ratio of Pd to the real part of Pin − Pout . one obtains Pd σ ωµ = e = Re{Pin − Pout } 2α β ω µσe 2ω 2 µ ε ′ 1 − cos δ e 2 = ω ε ′ tan δ e σe =1 (73) 2 cos δ e Chris Bishop 11/13/2001 . In (68) the sign of Pin is reversed when it is defined to make the sum of the output. z 0 . A.

The ratio of the dissipated power to real input power is found as A σ e E02 1 − e −2α z0 Pd 4α = A 2 β Re{Pin } E0 2 ωµ ( ) (75) Pd ω µ σe = 1 − e− 2 α z0 = 1 − e− 2 α z0 Re{Pin } 2 α β ( ) (76) If σ s = 0 then the more general equations (18). and (62) through (67) specialize as tan δ e = ε ′′ σ s ε ′′ + = ε′ ωε′ ε′ δe 2 (77) α = k sin (78) β = k cos δe 2 (79) (80) (81) ˆ E = E0 e −α z e − j β z x H= 1 ˆ E0 e −α z e − j ( β z +θ ) y η where η= µ = η e jθ ε (82) Pe = * E2 1 1 ˆ E × H ⋅ d s = ∫∫ 0 e −2α z e j θ z ⋅ d s 2 ∫∫ 2 S η S ( ) (83) Chris Bishop 11/13/2001 . Likewise. the dissipated (real) power is equal to the difference between the input and output real powers. one can show that 2ω (Wm + We ) =1 Im{Pin − Pout } (74) to prove that the stored (imaginary) power equals the difference in the input and output reactive powers. (51). (52).11 Thus.

and the fields go to zero. The assumptions that σ s ≠ 0 and ε ′′ = 0 do not yield any reduction of the equations. α and β become infinite. Chris Bishop 11/13/2001 . Advanced Engineering Electromagnetics.12 Pd = 2 1 1 2 − 2α z ∫∫∫ σ e E dv = 2 ∫∫∫ ωε ′′ E0 e dv 2 V V 2 1 1 µ H dv = ∫∫∫ ε E02 e −2α z dv ∫∫∫ 4 V 4 V 2 1 1 ε ′ E dv = ∫∫∫ ε ′ E 02 e − 2α z dv ∫∫∫ 4 V 4 V (84) Wm = (85) We = (86) As σ s → ∞ . Primary Source: Balanis.