Doubly Negative Metamaterials

Kirk T. McDonald
Joseph Henry Laboratories, Princeton University, Princeton, NJ 08544
(April 14, 2010; updated December 5, 2016)

1 Problem
In 1967 Veselago considered the possibility of materials with (relative) permittivity  and
(relative) permeability μ both negative [1]. Building on the long history of artifical dielectrics
for high-frequency applications (see, for example, [2, 3, 4]), in 1999 Pendry et al. proposed
a technique for fabrication of metamaterials with both negative  and negative μ within a
certain frequency range [5]. Such materials were first realized in the laboratory in 2000 [6].
An idealized model of the frequency dependence of the permittivity and the permeability
(both assumed to be linear and isotropic) is that
ω 2,p ω 2 − ω 2,1
(ω) = 1 ± ≈ , (1)
ω 2 − ω 2,0 + iΓ ω ω 2 − ω 2,0
and
ω 2μ,p ω 2 − ω 2μ,1
μ(ω) = 1 ± ≈ , (2)
ω 2 − ω 2μ,0 + iΓμ ω ω 2 − ω 2μ,0
where the time dependence is assumed to have the form e−iωt ,

ω1 = ω 20 ± ω 2p , (3)
ω p is equivalent to the plasma frequency of the medium, and ω 0 is a resonant frequency
of the medium that is associated with a small damping coefficient Γ which we neglect by
restricting the analysis to frequencies not too close to ω 0 or ω 1. In passive metamaterials the
± sign is negative, while for an active metamaterial [7] with an inverted population the sign
can be positive. In general, there is no relation between the frequencies ω ,j and ω μ,j , but it
suffices to suppose that either ω ,0 = ω μ,0, ω ,1 = ω μ,1 or ω ,0 = ω μ,1 , ω ,1 = ω μ,0 .
(a) Discuss the relation between the phase and group velocities, vp and vg , in a metama-
terial for waves of the form ei(kx−ωt) , where for negligible damping we can write (see
chap. 6, sec. 2 of [8] or sec. 85 of [9])
dω x̂ c x̂ c x̂
vg = x̂ = =± =± . (4)
dk dk/dω d[ωn]/dω n + ω dn/dω
with n = ±ck/ω being the index of refraction, and c is the speed of light in vacuum.
For what range of material parameters, if any, can the phase and group velocity be in
opposite directions and/or the index n be considered as negative?
(b) By consideration of the phases of planes waves at the interface between an ordinary
transparent medium and a metamaterial, deduce the form of Snell’s law for waves
incident on the interface at angle θi from within the ordinary medium, and refracted
into the metamaterial at angle θt to the normal.

1
2 Solution
2.1 Phase and Group Velocity
In a medium free of external charge and current, that can be characterized by linear, isotropic
(relative) permittivity  and permeability μ, Maxwell’s equations are (in Gaussian units)

1 ∂B 1 ∂D
∇ · D = 0, ∇×E = − , ∇ · B = 0, ∇×H = , (5)
c ∂t c ∂t
and the constitutive relations are

D = E, B = μH. (6)

These can be combined to give a wave equation for, say, the magnetic field H,

μ ∂ 2H
∇2 H = . (7)
c2 ∂t2
In general we consider plane waves of the form

H = H0 ei(k·x−ωt) , (8)

where H0 and k can be complex functions of the angular frequency ω (but not of time). The
approximations of the models (1)-(2) for metamaterials permit imply that the wave vector
k is purely real. In this case, eqs. (7) and (8) combine to give the dispersion relation

μω 2 n2 ω 2
k2 ≡ k 2 = ≡ (9)
c2 c2
where the scalar wave number k is defined to be positive, and n is the index of refraction.
For wave propagation, both  and μ must be positive, or both negative. Then,
√
μ ω nω
k= =± , (10)
c c
where the index n could be either positive or negative,
√
n = ± μ. (11)

We write the wave vector as k = k k̂ with k positive, so the waveform (8) becomes

H = H0 ei(k k̂·x−ωt), (12)

such that we identify the phase velocity vp as

ω c x̂ c
vp = k̂ = √ = ± k̂. (13)
k μ n

The phase-velocity vector vp is in the direction of the wave vector k = k k̂.

2
 Turning to the group velocity, we combining eqs. (4) and (10) to write the group velocity
of a packet of waves of the form (8) as1
1 c x̂ c x̂
vg = x̂ = √ =± . (14)
dk/dω d[ω μ]/dω d[ωn]/dω
For passive metamaterials the model forms (1)-(2) hold for ω 0 < ω 1 , as sketched on the
√
left in the two figures below. In this case d(ω μ)/dω is negative for ω 0 < ω < ω 1 where 
and μ are both negative. Thus, the phase and group velocities, eqs. (13) and (14), have the
opposite signs in a doubly negative, passive metamaterial [10].

√
We can use the notation n = − μ together with k = −nω/c to write the waveform (8)
as
H = H0ei(nωx/c−ωt) , (15)
for the case of a wave with group velocity in the +x-direction and phase velocity in the −x-
direction. This convention is often summarized by saying that a passive, doubly negative
metamaterial has a negative index of refraction.2
1

Care
 must be taken to avoid the mathematical oddity that one might write df/dω = d (−f)2 /dω =
2iπ
2(1/2) (−f)/(−f)√d(−f)/dω = −df/dω, which is a variant on the theme that while 1 = e we cannot
√
say that 1 = 1 = e2iπ = eiπ = −1.
2 √
Of course, it is also consistent to write n = μ and k = nω/c. Conversely, any medium could be
defined to have a negative index of refraction so long as we also write k = −nω/c.

3
 On the other hand, if active metamaterials can be constructed with behavior as shown in
√
the right figures above, then for ω 1 < ω < ω 0 , where both  and μ are negative, d(ω μ)/dω
is positive, and the group and phase velocities are in the same direction. In this case it would
√
be best to write simply n = μ.

2.1.1 Energy Density

If a medium could have negative  and negative μ at zero frequency, then the static energy
density,
E 2 + μH 2
u= , (16)
8π
would be negative, which is unreasonable for a passive material.
When the permittivity and permeability have frequency dependence, the energy density
takes on a more complicated form (due to Brillouin (1932), reviewed in chap. IV of [8]; see
also sec. 80 of [9]),  
1 d(ω)  2  d(μω)  2 
uω  = E + H . (17)
8π dω dω
The products ω and μω are illustrated for the models (1)-(2) in the figure below. We see
that for the passive metamaterial (ω 0 < ω 1) the energy density is positive (as was anticipated
by Veselago [1]). But for an active metamaterial the energy density is formally negative in
the region ω 1 < ω < ω 0 of doubly negative  and μ. Such behavior is nonclassical, but occurs
in gain media [11].3

3
In the realm of electromagnetic antennas there exists lore called the “Chu limit” [12, 13, 14, 15, 16],
which tacitly assumes that the density of electromagnetic energy in the “near zone” of the antenna is
positive. However, if the near zone were an active medium, perhaps a metamaterial, the Chu limit would
not necessarily hold [17]. It remains that to date no clear evasion of the Chu limit has been demonstrated.

4
2.1.2 “Left-Handed” Metamaterials
The fourth Maxwell equation (5) for a plane wave of form ei(k·x−ωt) indicates that

ω μω μ  √
k×E = B = H=± kH, or H=± k̂ × E, B= μ k̂ × E, (18)
c c  μ

using eq. (10), where the upper sign holds for ordinary media and active metamaterials, while
the lower sign holds for passive metamaterials. This result has led to the characterization
of passive metamaterials as left-handed media, in contrast to ordinary media, and active
metamaterials, for which k̂, E and H form a right-handed triad.4 According to the discussion
in sec. 2.1, only left-handed media should be characterized as having a negative index of
refraction.

2.1.3 Momentum Density

In 1903 Max Abraham noted [18] that the Poynting vector [19], which describes the flow of
energy in the electromagnetic field,
c
S= E × H, (19)
4π
when divided by c2 has the additional significance of being the density of momentum stored
in the electromagnetic field,

(A) E×H
pEM = (Abraham). (20)
4πc
Of course, D = E + 4πP and H = B − 4πM, where P and M are the densities of electric
and magnetic polarization, respectively.
In 1908 Hermann Minkowski gave an alternative derivation [20] that the electromagnetic-
momentum density is5

(M) D×B
pEM = (Minkowksi), (21)
4πc
and the debate over the merits of these two expressions continues to this day. Minkowski
died before adding to the debate, while Abraham published several times on it [22]. For
recent reviews, see [23, 24, 25].
As both fields D and H include aspects of the material medium, if any, we consider the
field-only momentum density to be simply
E×B
pEM = (field only). (22)
4πc
This momentum density is directed along the wave vector k, which is opposite to the group
velocity, in doubly negative metamaterials.
4
Note that k̂, E and B form a right-handed triad in all materials.
5
See also, for example, sec. 2.1 of [21].

5
 In general,
(A) (M)
pEM = μpEM , pEM = pEM , (23)
(A) (M)
such that in doubly negative metamaterials both pEM and pEM point in the direction of the
group velocity, but are equal only if  = μ.
In a dispersive medium with index n(ω) it is useful to introduce the quantity
c dk d(ωn) dn
ng = =c = =n+ω , (24)
vg dω dω dω
which is sometimes called the group-velocity index. This velocity is usually positive in a
passive medium, but is negative near optical resonances [11]. The emerging consensus [21,
23, 24, 25] is that the Abraham momentum density corresponds to the momentum of a photon
of angular frequency ω in a medium of group-velocity index ng being h̄ωvg /c2 = h̄ω v̂g /ng c,
and is sometimes call the kinetic momentum density. The Minkowski momentum density
corresponds to the momentum n2h̄ωvg /c2 = n2h̄ω v̂g /ng c of a photon of angular frequency
ω, and is sometimes called the pseudomomentum or the quasimomentum.6 The Abraham and
Minkowski momenta are always in the same direction for ordinary media and for doubly
negative metamaterials.
The momentum of a photon most often used in quantum theory is h̄k = nh̄ω k̂/c, which
is often called the canonical momentum. The canonical momentum is in the direction of the
wave vector k, and so has the same direction as the field-only momentum density (22). In
most media the Minkowski momentum is the same as the canonical momentum, but they
are opposite in a passive, doubly negative, nondispersive metamaterial [26].7
Which of these momenta is the most physically relevant depends on the details of the
experiment to measure the momentum.

2.2 Snell’s Law

We consider the case of an ordinary medium with 1 > 0, μ1 > 0 at x < 0, and a doubly
negative metamaterial with 2 < 0, μ2 < 0 at x > 0, as sketched in the figure on the next
page.
We also consider plane waves of the form E = E0ei(k·x−ωt). One condition on the incident,
reflected and transmitted waves is that the phase of all three waves must be the same at any
point (0, y, z) on the interface x = 0. This implies that
ki sin θi = kr sin θr = ±kt sin θt , (25)
where for the moment we suppose that the transmitted wave is “ordinary” in the sense that
the transmitted ray which passes through the origin lies in the first quadrant of the x-z
6
Some people prefer to define the Minkowksi momentum as h̄k, in which case it is identical to the
canonical momentum [23]. This view comes from consideration of the wave as a polariton, which is not
strictly a photon, as a polariton includes matter, charged or neutral, in its wave function. The momentum
of the polariton is n2 h̄ωvg /c2 , of which it is considered that the photon part is h̄k. I am not presently
persuaded of the merits of this decomposition of a quantum state.
7
So-called electrostatic waves can exist in media where D, B and H are zero , while the wave is associated
with a momentum density in the medium that is opposite to the direction of the wave vector [27].

6
plane, as shown in the left figure above for an active metamaterial. Then, the upper (lower)
sign in eq. (25) holds for an active (passive) metamaterial.

Since the incident and reflected waves are in the same medium the magnitudes of the
wave numbers ki and kr are the same (for waves of a given frequency ω),
√
1 μ1 ω n1 ω
ki = kr = = , (26)
c c
and hence
θr = θi . (27)
If we designate kt to be a positive quantity,
√
2 μ2 ω n2 ω
kt = =± , (28)
c c
where the index of refraction,
√
n2 = ± 2 μ2 , (29)
can be considered as negative for a passive metamaterial, as discussed in sec. 2.1. Then,
eq. (25) also indicates that
n1 sin θi = n2 sin θt , (30)
where n2 , and hence θ t as well, are negative for passive metamaterials. In the latter case,
the transmitted ray which passes through the origin lies in the second quadrant of the x-z
plane, as shown in the right figure above. This behavior is often called negative refraction.8,9
8
Thus, we have the option of retaining Snell’s law in the form n1 sin θi = n2 sin θt for medium 2 being a
passive metamaterial by considering n2 and θ t as both negative, with θ t defined to be positive in the first
quadrant of the x-z plane. Alternatively, we could consider n2 to be positive, and modify Snell’s law to be
n1 sin θ i = −n2 sin θt , where again θt is defined to be positive in the first quadrant of the x-z plane. And
yet another alternative is to write Snell’s law as n1 sin θ i = n2 sin θ t with n2 positive, but to define θ t to be
positive in the second quadrant of the x-z plane. Among these three alternatives the first is perhaps the
most pleasing.
9
Negative refraction in the sense considered here can also be occur for materials with positive Re(n) if
the losses are sufficiently high (Im(n) large) [28, 29].

7
Acknowledgment
The author thanks Sir John Pendry for discussions of this topic.

References
[1] V.G. Veselago, The Electrodynamics of Substances with Simultaneously Negative Values
of  and μ, Sov. Phys. Uspekhi 10, 509 (1968),
http://puhep1.princeton.edu/~mcdonald/examples/EM/veselago_spu_10_509_68.pdf

[2] W.E. Kock, Metal-Lens Antennas, Proc. IRE 34, 828 (1946),
http://physics.princeton.edu/~mcdonald/examples/EM/kock_pire_34_828_46.pdf

[3] J. Brown, Artificial Dielectrics Having Refractive Indices Less Than Unity, Proc. IEE
100, 51 (1953), http://physics.princeton.edu/~mcdonald/examples/EM/brown_piee_100_51_53.pdf

[4] W. Rotman, Plasma Simulation by Artificial Dielectrics and Parallel-Plate Media, IRE
Trans. Ant. Prop. 10, 82 (1962),
http://physics.princeton.edu/~mcdonald/examples/EM/rotman_iretap_10_82_62.pdf

[5] J.B. Pendry et al., Magnetism from Conductors and Enhanced Nonlinear Phenomena,
IEEE Trans. Micro. Theor. Tech. 47, 2075 (1999),
http://physics.princeton.edu/~mcdonald/examples/EM/pendry_ieeetmtt_47_2075_99.pdf

[6] D.R. Smith et al., Composite Medium with Simultaneously Negative Permeability and
Permittivity, Phys. Rev. Lett. 85, 4184 (2000),
http://physics.princeton.edu/~mcdonald/examples/EM/smith_prl_84_4184_00.pdf

[7] S.A. Ramakrishna and J.B. Pendry, Removal of absorption and increase in resolution
in a near-field lens via optical gain, Phys. Rev. B 67, 201101 (2003),
http://physics.princeton.edu/~mcdonald/examples/EM/ramakrishna_prb_67_201101_03.pdf

[8] L. Brillouin, Wave Propagation and Group Velocity (Academic Press, New York, 1960).

[9] L.D. Landau, E.M. Lifshitz and L.P. Pitaevskii, Electrodynamics of Continuous Media,
2nd ed. (Butterworth-Heinemann, Oxford, 1984).

[10] D.R. Smith and N. Kroll, Negative Refractive Index in Left-Handed Materials, Phys.
Rev. Lett. 85, 2933 (2000),
http://physics.princeton.edu/~mcdonald/examples/EM/smith_prl_85_2933_00.pdf

[11] K.T. McDonald, Negative Group Velocity (June 23, 2000),

http://physics.princeton.edu/~mcdonald/examples/negativegroupvelocity.pdf

[12] L.J. Chu, Physical Limitations of Omin-Directional Antennas, J. Appl. Phys. 19, 1163
(1948), http://physics.princeton.edu/~mcdonald/examples/EM/chu_jap_19_1163_48.pdf

8
[13] R.F. Harrington, Effect of Antenna Size on Gain, Bandwidth and Efficiency, J. Res.
Nat. bur. Stand. 64D, 1 (1960),
http://physics.princeton.edu/~mcdonald/examples/EM/harrington_jrnbs_64d_1_60.pdf

[14] H.A. Wheeler, Small Antennas, IEEE Trans. Ant. Prop. 23, 462 (1975),
http://physics.princeton.edu/~mcdonald/examples/EM/wheeler_ieeetap_23_462_75.pdf

[15] R.C. Hansen, Fundamental Limitations in Antennas, Proc. IEEE 69, 170 (1981),
http://physics.princeton.edu/~mcdonald/examples/EM/hansen_pieee_69_170_81.pdf

[16] J.S. McClean, A Re-Examination of the Fundamental Limits on the Radiation Q of

Electrically Small Antennas, IEEE Trans. Ant. Prop. 44, 672 (1996),
http://physics.princeton.edu/~mcdonald/examples/EM/mclean_ieeetap_44_672_96.pdf

[17] R.W. Ziolkowski and A. Erentok, At and below the Chu limit: passive and active broad
bandwidth metamaterial-based electrically small antennas, IET Microw. Ant. Prop. 1,
116 (2007), http://physics.princeton.edu/~mcdonald/examples/EM/ziolkowski_ietmap_1_116_07.pdf

[18] M. Abraham, Prinzipien der Dynamik des Elektrons, Ann. Phys. 10, 105 (1903),
http://physics.princeton.edu/~mcdonald/examples/EM/abraham_ap_10_105_03.pdf

[19] J.H. Poynting, On the Transfer of Energy in the Electromagnetic Field, Phil. Trans.
Roy. Soc. London 175, 343 (1884),
http://physics.princeton.edu/~mcdonald/examples/EM/poynting_ptrsl_175_343_84.pdf

[20] H. Minkowski, Die Grundgleichungen für die elektromagnetischen Vorgänge in bewegten

Körper, Nachr. Ges. Wiss. Göttingen 1, 53 (1908),
http://physics.princeton.edu/~mcdonald/examples/EM/minkowski_ngwg_53_08.pdf
http://physics.princeton.edu/~mcdonald/examples/EM/minkowski_ngwg_53_08_english.pdf

[21] P. Milonni and R.W. Boyd, Momentum of Light in a Dielectric Medium, Adv. Opt.
Phot. 2, 519 (2010), http://physics.princeton.edu/~mcdonald/examples/EM/milonni_aop_2_519_10.pdf

[22] M. Abraham, Zur Elelktrodynamik bewegten Körper, Rend. Circ. Matem. Palermo 28,
1 (1909), http://physics.princeton.edu/~mcdonald/examples/EM/abraham_rcmp_28_1_09.pdf
http://physics.princeton.edu/~mcdonald/examples/EM/abraham_rcmp_28_1_09_english.pdf
Sull’Elettrodinamica di Minkowski, Rend. Circ. Matem. Palermo 30, 33 (1910),
http://physics.princeton.edu/~mcdonald/examples/EM/abraham_rcmp_30_33_10.pdf
http://physics.princeton.edu/~mcdonald/examples/EM/abraham_rcmp_30_33_10_english.pdf
Zur Frage der Symmetrie des elektromagnetischen Spannungstensors, Z. Phys. 17, 537
(1914), http://physics.princeton.edu/~mcdonald/examples/EM/abraham_zp_15_537_14_english.pdf

[23] S.M. Barnett, Resolution of the Abraham-Minkowski Dilemma, Phys. Rev. Lett. 104,
070401 (2010), http://physics.princeton.edu/~mcdonald/examples/EM/barnett_prl_104_070401_10.pdf

[24] B.A. Kemp, Resolution of the Abraham-Minkowski debate: Implications for the elec-
tromagnetic wave theory of light in matter, J. Appl. Phys. 109, 111101 (2011),
http://physics.princeton.edu/~mcdonald/examples/EM/kemp_jap_109_111101_11.pdf

9
[25] D.J. Griffiths, Electromagnetic momentum, Am. J. Phys. 80, 7 (2012),
http://physics.princeton.edu/~mcdonald/examples/EM/griffiths_ajp_80_7_12.pdf

[26] M.G. Scullion and S.M. Barnett, Optical momentum in negative-index media, J. Mod.
Opt. 55, 2301 (2008), http://physics.princeton.edu/~mcdonald/examples/EM/scullion_jmo_55_2301_08.pdf

[27] K.T. McDonald, An Electrostatic Wave (July 28, 2002),

http://physics.princeton.edu/~mcdonald/examples/bernstein.pdf

[28] M. Sanz et al., Transmission measurements in wedge-shaped absorbing samples: An

experiment for observing negative refraction, Phys. Rev. E 67, 067601 (2003),
http://physics.princeton.edu/~mcdonald/examples/EM/sanz_pre_67_067601_03.pdf

[29] R.C. Hansen, Negative Refraction Without Negative Index, IEEE Trans. Ant. Prop.
56, 402 (2008), http://physics.princeton.edu/~mcdonald/examples/EM/hansen_ieeetap_56_402_08.pdf

10