Based on the dispersive feature of the dielectric function of noble metals and the wave vector conservation in physics, both the plasma effect and the complex refractive index, which are profoundly correlated to the complex dielectric function and permeability, have been studied and analyzed. The condition to induce a bulk or a surface plasma in the visible region will not be satisfied, and there will be one solution for the real and the imaginary parts of the refractive index, restricting it only to region I of the complex plane. The results given in this work will aid in understanding the properties of light transmission at the metal/dielectric interface as characterized by the law of refraction in nature.
I. INTRODUCTION
The phenomenon of light refracted at the interface of two media can be expressed by Snell’s law
[1]
:
where θ
_{1}
and θ
_{2}
are the incidence and the refraction angles, and n
_{1}
and n
_{2}
are the refractive indices of media 1 and 2, respectively. The refractive index will be a real and positive number for nonabsorbing media, as will be the incidence and the refraction angles. However, the refractive index ñ will be a complex number for optically absorbing materials, as will be the refraction angle
appearing in a more general formation of Snell’s law
[2
,
3]
:
Under certain conditions, a mathematical solution for the negative refraction index might happen as prescribed by Veselago
[4]
. Since then, great effort has been made to study the intrinsic physical origins of negative refraction with different approaches and assumptions, such as plasma resonances
[5]
, negative permeability μ
[4
,
6]
, a negative GoosHänchen shift
[7]
, fishnet magnetic effects
[8]
, and so on. Usually, the noble metals with strong optical absorption in the visible region are the fundamental materials used to make negative refraction occur in metamaterials. In the Drude region of the noble metals, a dielectric function with a negative real part is characteristic of “free” electrons, and in the interband region, the real part and the optical response of the dielectric function are determined by “bound” electrons. In terms of the complex plane, we have studied the plasma resonance effect and the refractive index deduced from the complex dielectric function and permeability under certain physical conditions. We hope the results with detailed discussions given in this work will further help the understanding of light propagation behavior at the metal/dielectric interface.
II. PLASMA RESONANCE EFFECT IN THE DRUDE REGION
The plasma resonance effect has been used to explain the origin of negative refraction at the metal/air interface and should be studied and discussed in more detail
[5
,
11]
. In the freeelectron approximation, the collective motion of the electron gas in the metal can be excited by an electric field at the plasma frequency ω
_{p}
to satisfy the condition D = 0 with ε = 0
[3
,
12
,
13]
. According to Maxwell’s equations:
where ω is the frequency of the electromagnetic field in oscillation and B is the magnetic field related to the vector H as B = μH. When the plasma resonance occurs at frequency ω
_{p}
and ε = 0, this leads to
which means that E
_{loc}
is a longitudinal field, which is not characteristic of an electromagnetic field of the photons interacting with electrons in the metal. The issue about whether the plasmons cannot be excited by the photons was mentioned by Wooten in his book
[12]
. Johnson and Christy
[14]
measured the dielectric functions of thermoevaporated thin noble metal films by using reflection and transmission methods to find that under the condition ε ≈ 0 (ε = ε
_{1}
+ iε
_{2}
, ε
_{1}
= 0 and ε
_{2}
is small) and in the 0.5 to 6.5eV photon energy region, the plasma resonance occurs only for Ag at about 3.75 eV which is lower than ℏω
_{p}
(≈ 9 eV), but not for Cu and Au. The zero of ε
_{1}
of Ag at about 3.75 eV is actually attributed to the onset of interband transitions in the region.
The plasmons are longitudinal excitations in the volume of an electron gas. As longitudinal excitations, they do not couple to transverse electromagnetic waves and are usually studied by using electron energy loss spectroscopy. When a surface or interface is present, surface plasmons exist if the dielectric functions have suitable properties. For a freeelectron gasvacuum interface, a surface plasmon is found at the frequency ε
_{1}
= 1, which is at
provided ε
_{2}
is not too large at this energy.
For the noble metals with a mean free time τ between collisions of the electrons in the near infrared and visible frequency regions, where ωτ >> 1 is satisfied, the dielectric function to be used is then the sum of ε
_{b}
and the Drude terms
where the term ε
_{b}
represents the polarization of the electrons in the core energy bands in response to the oscillating electric field.
For the noble metals, ε
_{b}
is large, and the surface plasmon condition, ε
_{1}
= 1, is shifted to energies well below
to a region where damping may be rather large. For Ag, the zero of ε
_{1}
is at about 3.75 eV where the value of ε
_{2}
is less than 0.5, indicating that in Ag there is a welldefined plasma resonance and that it is shifted in energy to far below the freeelectron gas value by the polarization associated with the interband transitions. In Cu and Au, there is too much damping from plasmon decay into interband excitations so that distinct peaks in the electron energy loss spectra, Im[1/(ε
_{1}
+ iε
_{2}
)], are not seen, nor are peaks in the surface energy loss spectra
[15

17]
.
As a medium with a positive dielectric constant ε
_{d}
contacts the metal, the electromagnetic field of only the ppolarized light (the field parallel to the incidence plane) with wave vector k
_{s}
along the metal/dielectric interface will stimulate a surface plasmon resonance at the surface plasma frequency ω
_{sp}
[13
,
15

17]
:
where k
_{0}
(k
_{0}
=ω/c) is the wave vector in vacuum. At the noblemetal/air interface, ε
_{d}
=1, ℏω
_{sp}
≈ 0.7 ℏω
_{p}
≈ 6.3 eV. At the noblemetal/dielectric interface, most dielectric media are transparent in the near infrared and visible regions
Dispersion relation of the photon energy E(ℏω) vs. the real (solid line) and imaginary (dashed line) parts of the complex surface wave vector ks (k_{s}= k_{s1} + ik_{s2}) in the Drude region. The sharp resonance peaks of the real and the imaginary surface wave vectors occur at the energy E = E_{sp} = Ep/√2≈ 6.2 eV, where (1 + ε_{1}) ≈ 0.
with a refractive index n
_{d}
in the 1.3  2.5 range, ε
_{d}
(ε
_{d}
= n
_{d}
^{2}
) in the 1.7  6.3 range, and ℏω
_{sp}
≈ (0.370.61) ℏω
_{p}
≈ 3.3  5.5 eV. The estimated values of the surfaceplasma frequency ω
_{sp}
under those conditions are in reasonable agreement
Dispersion curves of E(ℏω) vs. k_{s} from the dielectric functions measured in this and Johnson and Christy’s work ^{15)} for the noble metals: (a) Ag, (b) Au, and (c) Cu. The resonancelike features occurring at about 2.1 eV, 2.5 eV and 3.9 eV for Cu, Au and Ag, respectively, are not attributed to the surface plasma resonance, but to smaller values of the dielectric functions, which change dramatically in the region where the intraband and interband transitions cross near E_{g}.
with that shown in
Fig. 2
.
3
of Maier’s book
[13]
; thus, the surface plasmon will not be observed in the near infrared and visible regions where E < 3.1 eV.
In terms of the dispersion of the dielectric functions given in Eq. (5), the bulk plasma frequency ω
_{p}
of the noble metals has been precisely measured and is in good agreement with the theoretical prediction
[14
,
18

20]
. The characteristic of the surface plasma frequency ω
_{sp}
, however, has not been explicitly measured with confirmed comparison to the theoretical prediction for most metal/dielectric interface structures. For the simplest case with ε
_{d}
=1 at the metal/air interface, Eq. (6) can be rewritten as
If the imaginary part of the dielectric function is omitted in the Drude region near E
_{g}
where ε
_{1}
= 1, only possible for Ag, the wave vector k
_{s}
will be a pure infinite imaginary quantity, forbidding the propagation of the field along the interface.
Assuming omission of the core election polarization ε
_{b}
and the interband transitions, and inserting the Drude parameters for Cu (E
_{p}
= 8.82 eV, and τ = 2.6×10
^{14}
sec
[19]
into the dielectric functions in Eq. (5), the dispersion relation of the photon energy E(ℏω) vs. the complex surface wave vector k
_{s}
(k
_{s}
= k
_{s1}
+ik
_{s2}
) according to Eq. (7) is plotted in
Fig. 1
. Sharp resonance peaks of the real and imaginary surface wave vectors can be clearly seen to occur at nearly the same frequency ω
_{sp}
or energy
where (1 + ε
_{1}
) ≈ 0, but there is no surface plasma resonance in the near infrared and visible Drude region where E < E
_{sp}
. For the attenuated field of the surface wave which will propagate a distance d in absorptive material, the intensity of the wave will decay to exp(2k
_{s2}
d). The high value (~ 4×10
^{5}
cm
^{1}
) of the imaginary wave vector k
_{s2}
means that the surface wave along the interface can propagate only a very short distance d of about 12.5 nm at an energy E = E
_{sp}
.
To consider more practically the situation in which both intraband and interband transitions occur in the region E < E
_{p}
, we used the experimental data for the dielectric functions measured in this and Johnson and Christy’s work
[14]
for the noble metals Cu, Au and Ag to calculate the dispersion curves, E(ℏω) vs. k
_{s}
, with the results shown in
Fig. 2
. There are indeed some resonancelike features at about 2.1 eV, 2.5 eV and 3.9 eV for Cu, Au and Ag, respectively. In terms of data analysis, it can be seen clearly that these resonancelike features are actually not attributed to the surface plasma resonance but to the dip and smaller values of the dielectric functions, which change dramatically in the region where the intraband and the interband transitions near E
_{g}
cross, as discussed above. The resonancelike feature for Cu gets much weaker due to the onset of interband transitions closer to the tail of the intraband transition in the region where the magnitude of the dielectric function of Cu is large.
Turbadar first observed the resonancelike dip in the internal reflection of ppolarized light in the visible region for a structure in which thin metal films were deposited on the bottom of a glass prism. He did not attribute the phenomenon to a surface plasma resonance because the feature could be well explained by using conventional film equations
[21
,
22]
. Later, Otto
[23]
and Kretschmann
[24]
carried out similar measurements and introduced the plasma polarization concept to explain the phenomenon with resonancelike features that changed with the wavelength and the incidence angle. These were radiative surface plasmons
[17]
, excited by the longitudinal component of the evanescent electromagnetic wave at the interface. Their frequencies were below that of the nonradiative surface plasmon mentioned earlier.
III. EFFECT OF NEGATIVE PERMEABILITY
For the complex refractive index ñ(ñ=n+ik, where n and k are the refractive index and the extinction coefficient, respectively) and the complex refraction angle
Eq. (2) can be modified as
where
and the effective refraction index n
_{2e}
and refraction angle θ
_{2e}
are real numbers that can be quantitatively determined along with their signs, but they usually will be a complicated expression in terms of n
_{1}
, θ
_{1}
, ñ
_{2}
, the wavelength λ and other physical parameters, which needs to be further studied and verified
[3
,
25]
. It is obvious to see in Eqs. (1), (2), (8) and (9) that n
_{2e}
= n
_{2}
will be satisfied only under the condition k
_{2}
= 0 for transparent materials; otherwise, n
_{2e}
≠ n
_{2}
and n
_{2e}
≠ ñ
_{2}
.
To explain the mechanism of negative refraction for artificial metamaterials, we have assumed a magnetic permeability, μ ≠ 1, μ < 0, and even a complex μ (μ = μ
_{1}
+ iμ
_{2}
), though it is not exactly measured or confirmed by experiment
[4
,
26

30]
. If a complex permeability is assumed, the complex refractive index ñ will be a solution of that:
where
Referring to the complex planes of ε
_{e}
and ñ shown in
Fig. 3
, the situation in which n < 0 and k > 0 will be satisfied because ε
_{e}
is restricted to regions III and IV of the lower half of the complex plane, i.e., the region where ε
_{e2}
= ε
_{1}
μ
_{2}
+ ε
_{2}
μ
_{1}
≤ 0 and 2π ≥
^{θ}
_{εe}
≥ π. The positive rate of energy absorption of the electromagnetic field interacting with metals requires that ε
_{2}
≥ 0 and μ
_{2}
≥ 0
[31]
.
In region IV, ε
_{e1}
= ε
_{1}
μ
_{1}
 ε
_{2}
μ
_{2}
≥ 0 and ε
_{e2}
= ε
_{1}
μ
_{2}
+ ε
_{2}
μ
_{1}
≤ 0, implying that ε
_{1}
and μ
_{1}
will have the same sign, i.e., either ε
_{1}
< 0 and μ
_{1}
<0, or ε
_{1}
> 0 and μ
_{1}
> 0 for the condition of ε
_{e1}
≥ 0, but ε
_{e2}
≤ 0 will not be satisfied for the case in which ε
_{1}
> 0 and μ
_{1}
> 0. When
^{θ}
_{εe}
continuously changes to the region in which
^{θ}
_{εe}
= 2π, θ
_{n}
= π and ε
_{e2}
= ε
_{1}
μ
_{2}
+ ε
_{2}
μ
_{1}
= 0, k moves to 0 and ñ (ñ = n,) becomes a pure negative real number
[8]
, but ε
_{e2}
= ε
_{1}
μ
_{2}
+ ε
_{2}
μ
_{1}
= 0, which means that ε
_{1}
and μ
_{1}
will have opposite signs, contradicting the condition that ε
_{e1}
= ε
_{1}
μ
_{1}
 ε
_{2}
μ
_{2}
> 0 in which ε
_{1}
and μ
_{1}
should have the same sign. This makes a pure negative refractive index (ñ = ñ) invalid even for the situation in which the complex magnetic permeability μ is introduced as an assumption.
The reflectivity R under the normal incidence condition is defined in optics as that
[3
,
12]
:
The complex planes of ε_{e} (ε_{e} = ε_{e}e^{iθεe}) and ñ (ñ = √ε_{e}= ne^{iθn}) showing that θn = ^{θ}_{εe}/2 and 2π ≥ ^{θ}_{εe}≥ π for the situation in which n < 0, k > 0, and ε_{e} is restricted to regions IIIIV, the lower part of the complex plane.
In regions I and II, as n changes from negative to positive by passing the zero point at which n = 0, implying that R = 1 which is not valid and has not yet observed in experiment for the optically absorbing and passive materials and structures in the nature.
In region III,
^{θ}
_{εe}
= 3π/2, θ
_{n}
= 3π/4, ε
_{e1}
= 0 and ε
_{e2}
< 0 lead to k = n and ε
_{1}
/ε
_{2}
= μ
_{2}
/μ
_{1}
, implying that ε
_{1}
and μ
_{1}
have the same sign, i.e., ε
_{1}
< 0 and μ
_{1}
< 0 for the noble metals. Because ε
_{1}
/ε
_{2}
 >> 1 for the noble metals in the near infrared and visible parts of the Drude region where E < E
_{g}
, μ
_{2}
>> μ
_{1}
 means a very high absorption of the magnetic field in the metal. As
^{θ}
_{εe}
= π, θ
_{n}
= π/2 and ñ = ik, the refractive index will be a pure imaginary number, forbidding propagation of the electromagnetic field in the metal, as mentioned above. At the same time, as
^{θ}
_{εe}
changes to the position where
^{θ}
_{εe}
= π and ε
_{e2}
= 0, ε
_{1}
and μ
_{1}
will have opposite signs, and μ
_{1}
changes sign from negative to positive due to the negative ε
_{1}
in the near infrared and visible regions. According to the dispersion of the dielectric function and permeability, therefore, although mathematically it allows the phase angle θ
_{n}
of ñ to be in the range 0 to π, by introducing the complex permeability μ, the invalidation of the value at which θ
_{n}
equals π/2 and π will make the refractive index ñ physically restricted to only region I of the complex plane where n > 0 and k > 0, as seen in
Fig. 3
.
The magnetic permeability μ in solids arises from micro magnetic dipole moments M responding to an alternating electromagnetic field according to the relations M = χ
_{m}
H and μ = 1 + 4πχ
_{m}
, where χ
_{m}
is the magnetic susceptibility. In the near infrared and visible regions with high optical frequencies, micromagnetic dipole moments cannot follow the rapid oscillations of the electromagnetic field, resulting in the physical restriction that χ
_{m}
= 0 and μ = 1
[3
,
12
,
32]
. No serious violation of the restriction μ = 1 has been found in numerous experiments that measured the dielectric function and the refractive index for many materials, including metals, in a wide photon energy range
[33]
.
IV. CONCLUSION
In this work, we have studied and analyzed the plasma resonance effect and the complex refractive index, which are profoundly correlated to the complex dielectric function and permeability. Due to the dispersive feature of the dielectric function of noble metals and the wavevector conservation required in physics, it may not be possible for either a bulk or a surface plasma to be induced by photons in the visible region. Mathematically, there are possible two solutions for the refractive index deduced from the square root of the dielectric function and permeability. Physically, however, there will be only one solution for the real and the imaginary parts of the refractive index, restricting it to region I of the complex plane. The results given in this work will be helpful to understand the light refraction phenomena observed in metalbased materials and the structures over the broad frequency region.
Acknowledgements
This work was supported by the National Science Foundation (NSF) project of China under contract number: #60938004 and by the Science and Technology Commission of Shanghai Municipality of China (Grant No. 08DJ1400302).
Shirley J. W.
(1951)
“An early experimental determination of Snell’s law”
Am. J. Phys.
19
507 
508
Kovalenko S. A.
(2001)
“DescartesSnell law of refraction with absorption”
Semiconductor Physics, Quantum Electronics & Optoelectronics
4
214 
218
Born M.
,
Wolf E.
1993
Principles of Optics
Pergamon
Oxford
Veselago V. G.
(1968)
“The electrodynamics of substances with simultaneously negative values of ε andμ”
Sov. Phys. Uspekhi
10
509 
514
Shin H.
,
Fan S.
(2006)
“Allangle negative refraction for surface plasmon waves using a metaldielectricmetal structure”
Phys. Rev. lett.
96
073907 
Depine R. A.
,
Lakhtakia A.
(2004)
“A new condition to identify isotropic dielectricmagnetic materials displaying negative phase velocity”
Microwave Opt. Technol. Lett.
41
315 
316
Dolling G.
,
Klein M. W.
,
Wegener M.
,
Schadle A.
,
Kettner B.
,
Burger S.
,
Linden S.
(2007)
“Negative beam displacements from negativeindex photonic metamaterials”
Opt. Express
15
14219 
14227
Pendry J. B.
(2000)
“Negative refraction makes a perfect lens”
Phys. Rev. Lett.
85
3966 
3969
Meyrath T. P.
,
Zentgraf T.
,
Giessen H.
(2007)
“Lorentz model for metamaterials: optical frequency resonance circuits”
Phys. Rev. B
75
205102 
Valentine J.
,
Zhang S.
,
Zentgraf T.
,
UlinAvila E.
,
Genov D. A.
,
Bartal G.
,
Zhang X.
(2008)
“Threedimensional optical metamaterial with a negative refractive index”
Nature
455
376 
379
Wegener M.
,
Dolling G.
,
Linden S.
(2007)
“Plasmonics: backward waves moving forward”
Nature Materials
6
475 
476
Wooten F.
1972
Optical Properties of Solids
Academic Press
New York and London
Maier S. A.
2007
Plasmonics: Fundamentals and Applications
Springer
New York, USA
Johnson P. B.
,
Christy R. W.
(1972)
“Optical constants of the noble metals”
Phys. Rev. B
6
4370 
4379
Raether H.
1980
Excitation of Plasmons and Interband Transitions by Electrons, Vol. 88 of Springer Tracts in Modern Physics
Springer Verlag
Berlin, Germany
Raether H.
1988
Surface Plasmons on Smooth and Rough Surfaces and on Gratings, Vol. 111 of Springer Tracts in Modern Physics
Springer Verlag
Berlin, Germany
Forstmann F.
,
Gerhardts R. R.
1986
Metal Optics, Vol. 109 of Springer Tracts in Modern Physics
Springer Verlag
Berlin, Germany
Aspnes D. E.
,
Kinsbron E.
,
Bacon D. D.
(1980)
“Optical properties of Au: sample effects”
Phys. Rev. B
21
3290 
3299
Parkins G. R.
,
Lawrence W. E.
,
Christy R. W.
(1981)
“Intraband optical conductivity σ(ω,T) of Cu, Ag, and Au: contribution from electronelectron scattering”
Phys. Rev. B
23
6408 
6416
Chen L. Y.
,
Lynch D. W.
(1987)
“Effect of liquids on the Drude dielectric function of Ag and Au films”
Phys. Rev. B
36
1425 
1431
Turbadar T.
(1959)
“Complete absorption of light by thin metal films”
Proc. Phys. Soc. Lond.
73
41 
44
Turbadar T.
(1964)
“Complete absorption of plane polarized light by thin metallic films”
J. Mod. Opt.
11
207 
210
Otto A.
(1968)
“Excitation of nonradiative surface plasma waves in silver by the method of frustrated total reflection”
Z. Physik
216
398 
410
Kretschmann E.
(1971)
“Die bestimmung optischer konstanten von metallen durch anregung von oberflächenplasmaschwingungen”
Z. Physik
241
313 
324
GarciaPomar J. L.
,
NietoVesperinas M.
(2004)
“Transmission study of prisms and slabs of lossy negative index media”
Opt. Express
12
2081 
2095
Shalaev V. M.
(2007)
“Optcal negativeindex metamaterials”
Nature Photonics
1
41 
48
McCall M. W.
,
Lakhtakia A.
,
Weiglhofer W. S.
(2002)
“The negative index of refraction demystified”
Eur. J. Phys.
23
353 
359
Depine R. A.
,
Lakhtakia A.
(2004)
“A simple and efficient approach to train artificial neural networks using a genetic algorithm to calculate the resonant frequency of an RMA on thick substrate”
Microwave Opt. Technol. Lett.
41
315 
316
Kildishev A. V.
,
Drachev V. P.
,
Chettiar U. K.
,
Werner D.
,
Kwon D. H.
,
Shalaev V. M.
(2006)
Wei J.
,
Xiao M.
(2007)
“Electric and magnetic losses and gains in determining the sign of refractive index”
Opt. Commun.
270
455 
464
Jackson J. D.
1999
Classical Electrodynamics
John Wiley & Sons, Inc.
New York, USA
Landau L. D.
,
Lifshitz E. M.
,
Pitaevskii L. P.
1984
Electrodynamics of Continuous Media
Pergamon
New York, USA
Palik E. D.
1985
Handbook of Optical Constants of Solids
Academic Press
London, UK