1: %\documentclass[preprint,prb,showpacs,aps]{revtex4}
2:
3:
4: \documentclass[twocolumn,showpacs,prb,aps]{revtex4}
5: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
6: \usepackage{amsmath}
7: \usepackage{graphicx}
8:
9: \begin{document}
10:
11: \title{All-angle zero reflection at metamaterial surfaces}
12: \author{Xin Li,$^1$ Zixian Liang,$^2$ Xiaohan Liu,$^{1,3}$ Xunya Jiang,$^2$
13: and Jian Zi$^{1,3}$\footnote{%
14: Electronic mail: jzi@fudan.edu.cn}}
15: \affiliation{$^1$Department of Physics and Surface Physics Laboratory, Fudan University,
16: Shanghai 200433, People's Republic of China\\
17: $^2$Shanghai Institute of Microsystem and Information Technology, CAS,
18: Shanghai 200050, People's Republic of China\\
19: $^3$Laboratory of Advanced Materials, Fudan University, Shanghai 200433,
20: People's Republic of China}
21: \date{\today}
22:
23: \begin{abstract}
24: The authors study theoretically reflection on the surface of a metamaterial
25: with a hyperbolic dispersion. It is found that reflection is strongly
26: dependent on how the surface is terminated with respect to the asymptote of
27: the hyperbolic dispersion. For a surface terminated normally to the
28: asymptote, zero reflection occurs for all incident angles. It is exemplified
29: by a metamaterial made of a periodic metal-dielectric layered structure with
30: its surface properly cut through numerical simulations.
31: \end{abstract}
32:
33: \pacs{42.25.Gy, 78.67.Pt, 78.20.Ci, 41.20.Jb}
34: \maketitle
35:
36: Metamaterials are artificially designed composites consisting of periodic
37: subwavelength structures. The optical response of metamaterials originates
38: from their structures instead from their compositions, leading to many
39: unusual optical properties that do not occur in nature. For instance,
40: metamaterials with a negative refractive index can produce negative
41: refraction\cite{she:01,par:03,hou:03} and superlensing.\cite%
42: {pen:00,grb:04,fan:05} In contrast to conventional materials, the energy
43: transport through negative-refractive-index metamaterials is in a direction
44: opposite to the phase direction, giving rise to reversed Doppler effects and
45: inverted Cherenkov cone.\cite{ves:68,pen:06a} Metamaterials can also be used
46: to construct invisible cloak.\cite{pen:06,sch:06}
47:
48: Reflection is a wave phenomenon occurring for waves impinging upon a
49: surface. Our common understanding is that reflection is inevitable although
50: it can be eliminated at some special incident angles, e.g., Brewster's
51: angles. In this Letter, we show theoretically that all-angle zero reflection
52: can occur at the surface of a metamaterial with a hyperbolic dispersion. It
53: is exemplified by a metamaterial made of a periodic metal-dielectric layered
54: structure through numerical simulations.
55:
56: \begin{figure}[htbp]
57: \centerline{\includegraphics[angle=0,width=7cm]{fig1.eps}}
58: \caption{(Color online) Equi-frequency surface analysis of
59: reflection on surfaces of metamaterials with (a) an elliptic and
60: (b-d) a hyperbolic dispersion. Surfaces (dash-dotted lines) lie in
61: the horizontal plane. For metamaterials with the hyperbolic
62: dispersion, surfaces are cut along the principal axis (b),
63: perpendicularly to the asymptote (dashed lines) of the hyperbolic
64: dispersion (d), and obliquely to both the principal axis and
65: asymptote (c). Black (grey) thick arrows denote the incident
66: (reflected) wave vector. Thin arrows indicate the direction of the
67: group velocity. Dotted lines illustrate the conservation of the
68: in-plane wave vectors. Note that metamaterials occupy the half-space
69: below the horizontal line and waves are incident from the
70: metamaterial side. } \label{fig1}
71: \end{figure}
72:
73: For any reflection on a surface, the spatial and time variation of
74: all fields must be the same at the surface. As a result, the
75: in-plane wave vector of an incident wave should be equal to that of
76: the reflected one, independent of the nature of the boundary
77: conditions.\cite{jac:99} For homogeneous media, this leads to the
78: fact that the incident angle is equal to the reflected angle. For
79: anisotropic media, however, incident and reflected angles may not be
80: the same. For either isotropic media or anisotropic media with an
81: elliptic dispersion, reflection always exists due to the fact that
82: the conservation of in-plane wave vectors of incident and reflected
83: waves can be always satisfied, regardless of the cutting directions
84: of surfaces, as shown schematically in Fig. \ref{fig1}. For an
85: anisotropic medium with a hyperbolic dispersion, reflection is
86: sensitive to the surface termination direction, namely, the surface
87: orientation with respect to the asymptote of the hyperbolic
88: dispersion. For surfaces terminated along the principal axes,
89: reflection always exists with equal incident and reflected angles.
90: For surfaces oriented obliquely to both the asymptotes and principal
91: axes, the reflected wave vector still exists with the reflected
92: angle different from the incident angle. If the surface is cut
93: perpendicularly to the asymptote, however, we cannot find reflected
94: wave vector for any given incident wave vector. In this situation,
95: zero reflection is expected for all incident angles as to be shown
96: later. It should be noted that the energy flow direction is the same
97: as that of the group velocity, defined by $\mathbf{v}_{g}=\nabla
98: \omega (\mathbf{k})$, where $k$ is the wave vector and $\omega
99: (\mathbf{k})$ is the dispersion relation. From its definition, the
100: group velocity direction is perpendicular
101: to the equi-frequency surface and points to the direction along which $%
102: \omega (\mathbf{k})$ is increasing.
103:
104: A qualitative analysis of all-angle zero reflection was given
105: hereinbefore. In the following, we would like to give a rigorous
106: proof. We consider an anisotropic metamaterial whose permittivity
107: tensor and permeability tensor are both diagonalizable.\cite{smi:03}
108: To simply the proceeding analysis, we assume that the metamaterial
109: is nonmagnetic, namely, the diagonal elements of the permeability
110: tensor are all equal to 1. Without loss of generality, we assume in
111: our analysis a plane wave with the magnetic field polarized
112: along the $y$ direction with the form $\mathbf{H}=H_{0}\widehat{\mathbf{y}}%
113: \exp \left[ i(k_{x}x+k_{z}z)-\omega t\right] $, where $\widehat{\mathbf{y}}$
114: is the unit vector along the $y$ direction. If we choose a Cartesian
115: coordinate system $x$-$z$ with its axes along the principal axes of the
116: metamaterial, this plane wave satisfies the following dispersion relation%
117: \begin{equation}
118: \frac{k_{x}^{2}}{\varepsilon _{z}}+\frac{k_{z}^{2}}{\varepsilon _{x}}=\frac{%
119: \omega ^{2}}{c^{2}}, \label{dis}
120: \end{equation}%
121: where $\varepsilon _{x}$ and $\varepsilon _{z}$ are the diagonal elements of
122: the permittivity tensor. For $\varepsilon _{x}$ and $\varepsilon _{z}$ with
123: opposite signs, the corresponding dispersion is hyperbolic. If we choose a
124: new Cartesian coordinate system $x^{\prime }$-$z^{\prime }$ with the same
125: origin, the permittivity tensor is no longer diagonal and is transformed to%
126: \begin{equation}
127: \left[
128: \begin{array}{lr}
129: \varepsilon _{x^{\prime }x^{\prime }} & \varepsilon _{x^{\prime }z^{\prime }}
130: \\
131: \varepsilon _{z^{\prime }x^{\prime }} & \varepsilon _{z^{\prime }z^{\prime }}%
132: \end{array}%
133: \right] =\left[
134: \begin{array}{lr}
135: c^{2}\varepsilon _{x}+s^{2}\varepsilon _{z} & cs(\varepsilon
136: _{x}-\varepsilon _{z}) \\
137: cs(\varepsilon _{x}-\varepsilon _{z}) & s^{2}\varepsilon
138: _{x}+c^{2}\varepsilon _{z}%
139: \end{array}%
140: \right] .
141: \end{equation}%
142: Here, the parameters $c$ and $s$ are given by
143: \begin{equation}
144: c=\cos ^{2}\theta ,\text{ \ }s=\sin ^{2}\theta ,
145: \end{equation}
146: where $\theta $ is the angle between the $x^{\prime }$ and $x$ axes. The
147: dispersion in the new coordinate system becomes accordingly%
148: \begin{equation}
149: \frac{\varepsilon _{x^{\prime }x^{\prime }}^{2}k_{x^{\prime
150: }}^{2}+\varepsilon _{z^{\prime }z^{\prime }}^{2}k_{z^{\prime
151: }}^{2}+2\varepsilon _{x^{\prime }z^{\prime }}k_{x^{\prime }}k_{z^{\prime }}}{%
152: \varepsilon _{x^{\prime }x^{\prime }}\varepsilon _{z^{\prime }z^{\prime
153: }}-\varepsilon _{x^{\prime }z^{\prime }}^{2}}=\frac{\omega ^{2}}{c^{2}}.
154: \label{dis2}
155: \end{equation}
156:
157: For a metamaterial with a hyperbolic dispersion, reflection is strongly
158: dependent on the cutting direction of the surface. Without loss of
159: generality, we assume that the metamaterial surface is terminated normally
160: to the $z^{\prime }$ axis, i.e., in the $x^{\prime }y^{\prime }$ plane.
161: Consequently, $k_{x^{\prime }}$ is the in-plane wave vector of an plane wave
162: and $k_{z^{\prime }}$ is the perpendicular component. For a given in-plane
163: wave vector $k_{x^{\prime }}$ of an incident plane wave, we can always find
164: a real solution of $k_{z^{\prime }}$ for the reflected wave from the
165: conservation of the in-plane wave vector, if the $x^{\prime }$ axis is not
166: perpendicular to the asymptote of the hyperbolic dispersion. For the $%
167: x^{\prime }$ axis just perpendicular to the asymptote, no real solution of $%
168: k_{z^{\prime }}$ for the reflected wave can be found. In this case, $%
169: k_{z^{\prime }}$ should be a complex number possessing both a real and an
170: imaginary part. Zero reflection is expected as can be confirmed by
171: calculating the normal component of the reflected Poynting vector with
172: respect to the surface, defined in cgs units by%
173: \begin{equation}
174: S_{r\perp }=\frac{4\pi }{c}\mathrm{Re}\left( \mathbf{E}_{r}\times \mathbf{H}%
175: _{r}^{\ast }\right) _{\perp }=\frac{4\pi }{c}\mathrm{Re}\left( E_{rx^{\prime
176: }}H_{ry^{\prime }}^{\ast }\right) ,
177: \end{equation}%
178: where $E_{r}$ and $H_{r}$ are the reflected electric and magnetic fields
179: with their components related each other by
180: \begin{equation}
181: E_{rx^{\prime }}=-\frac{c}{\omega }\frac{\varepsilon _{z^{\prime }z^{\prime
182: }}k_{z^{\prime }}+\varepsilon _{x^{\prime }z^{\prime }}k_{x^{\prime }}}{%
183: \varepsilon _{x^{\prime }x^{\prime }}\varepsilon _{z^{\prime }z^{\prime
184: }}-\varepsilon _{x^{\prime }z^{\prime }}^{2}}H_{ry^{\prime }}.
185: \end{equation}%
186: Suppose a complex $k_{z^{\prime }}$ for the reflected wave and substitute it
187: into Eq. (\ref{dis2}) we can obtain
188: \begin{equation}
189: \varepsilon _{z^{\prime }z^{\prime }}\mathrm{Re}(k_{z^{\prime
190: }})+\varepsilon _{x^{\prime }z^{\prime }}k_{x^{\prime }}=0.
191: \end{equation}%
192: This immediately leads to the fact that $E_{rx^{\prime }}$ does not possess
193: a real part, leading to $S_{r\perp }=0$. We can thus conclude that the
194: reflected wave does not carry any energy along the surface normal, implying
195: zero reflection.
196:
197: One feasible realization of a metamaterial with a hyperbolic dispersion is
198: to adopt a periodic metal-dielectric layered structure.\cite%
199: {ram:03,bel:06,sal:06,liu:07} In the long wavelength limit (the period is
200: much smaller than the operating wavelength), this periodic metal-dielectric
201: layered structure can be viewed as a effective anisotropic metamaterial
202: with the permittivity tensor given by%
203: \begin{equation}
204: \left(
205: \begin{array}{ccc}
206: \varepsilon _{x} & 0 & 0 \\
207: 0 & \varepsilon _{x} & 0 \\
208: 0 & 0 & \varepsilon _{z}%
209: \end{array}%
210: \right) .
211: \end{equation}%
212: For $p$-polarized waves, $\varepsilon _{x}$ and $\varepsilon _{z}$
213: are related to the parameters of the constituents
214: by\cite{ram:03,ber:78}
215: \begin{subequations}
216: \begin{eqnarray}
217: \varepsilon _{x} &=&\frac{\varepsilon _{1}d_{1}+\varepsilon _{2}d_{2}}{d},%
218: \text{ } \\
219: \varepsilon _{z} &=&\frac{\varepsilon _{1}\varepsilon _{2}d}{\varepsilon
220: _{1}d_{2}+\varepsilon _{2}d_{1}},
221: \end{eqnarray}%
222: \end{subequations}
223: where $\varepsilon _{1,2}$ is the dielectric constant of the constituents, $%
224: d_{1,2}$ is the thickness, and $d=d_{1}+d_{2}$ is the period. If we adopt a
225: Drude model\cite{ash:76} to describe the dielectric constant of the metal
226: constituent, namely, $\varepsilon (\omega )=1-\omega _{p}^{2}/\omega ^{2},$
227: where $\omega _{p}$ is the plasma frequency of the metal, it can be easily
228: shown that $\varepsilon _{x}$ and $\varepsilon _{z}$ can have opposite signs
229: at certain frequencies for a proper choice of the thickness parameters. This
230: leads to a hyperbolic dispersion for the periodic metal-dielectric layered
231: structure.
232:
233: To illustrate all-angle zero reflection on the surface of a periodic
234: metal-dielectric layered structure, we carry out finite-difference
235: time-domain (FDTD) simulations\cite{taf:95} with perfectly matched layer
236: boundary conditions,\cite{ber:94} shown in Fig. \ref{fig2}. Without loss of
237: generality, the periodic metal-dielectric layered structure is assumed to
238: made from Ag and a dielectric with a dielectric constant of 2.231. The
239: thickness of the Ag layer is 10 nm and that of the dielectric layer is 50
240: nm. In our simulations, a $p$-polarized Gaussian beam is used. Its
241: wavelength is 756 nm, which is much larger than the period (60 nm) of the
242: periodic metal-dielectric layered structure, justifying our effective medium
243: approximation. An experimental value\cite{joh:72} of the refractive index $%
244: n=0.03+i5.242$ at 756 nm for Ag is used. These parameters result in Re$%
245: (\varepsilon _{x})=2.72$ and Re$(\varepsilon _{z})=-2.72$, leading to a
246: rectangularly hyperbolic dispersion for the periodic metal-dielectric
247: layered structure.
248:
249: \begin{figure}[tbp]
250: \centerline{\includegraphics[angle=0,width=7cm]{fig2.eps}}
251: \caption{(Color online) FDTD simulations of the magnetic field
252: distributions for a $p$-polarized Gaussian beam launched from a
253: periodic metal-dielectric layered structure upon an interface
254: between the structure (grey area) and a dielectric medium (white
255: area) with a dielectric constant of 2.5.} \label{fig2}
256: \end{figure}
257:
258: The terminated surface of the periodic metal-dielectric layered
259: structure is perpendicular to one of the asymptote of the hyperbolic
260: dispersion, namely, in an angle of 45$^{\circ }$ with respect to the
261: periodic direction. The incident Gaussian beam forms an angle of
262: 45$^{\circ }$ with respect to the surface normal and is
263: perpendicular to the periodic direction. It is obvious from the FDTD
264: simulations that no reflection occurs at the surface. It should be
265: noted that zero reflection does not depend on the incident angle.
266: Simulations for other incident angles are also conducted and zero
267: reflection is always found, manifesting all-angle zero reflection.
268:
269: In conclusion, the reflection on the surface of a metamaterial with a
270: hyperbolic dispersion is studied theoretically. For the metamaterial surface
271: terminated obliquely to the asymptote of the hyperbolic dispersion,
272: reflection is inevitable. However, all-angle zero reflection can occur if
273: the surface is cut perpendicularly to the asymptote. We show this by a
274: rigorous proof that for any incident angle the surface normal component of
275: the Poynting vector of the reflected wave does not possess a real part,
276: implying zero energy flow along the surface normal. Numerical simulations of
277: a periodic metal-dielectric layered structure that is cut properly affirm
278: unambiguously our theoretical prediction of all-angle zero reflection.
279: Metamaterials with all-angle zero reflection at their surfaces could be
280: exploited in many applications to eliminate undesirable reflection.
281:
282: This work was supported by the 973 Program (grant nos. 2007CB613200 and
283: 2006CB921700). The research of X.H.L and J.Z is further supported by NSFC
284: and Shanghai Science and Technology Commission is also acknowledged.
285:
286:
287: \begin{thebibliography}{99}
288: \bibitem{she:01} R. A. Shelby, D. R. Smith, and S. Schultz, Science \textbf{%
289: 292}, 77 (2001).
290:
291: \bibitem{par:03} C. G. Parazzoli, R. B. Greegor, K. Li, B. E. C. Koltenbah,
292: and M. Tanielian, Phys. Rev. Lett. \textbf{90}, 107401 (2003).
293:
294: \bibitem{hou:03} A. A. Houck, J. B. Brock, and I. L. Chuang, Phys. Rev.
295: Lett. \textbf{90}, 137401 (2003).
296:
297: \bibitem{pen:00} J. B. Pendry, Phys. Rev. Lett. \textbf{85}, 3966 (2000).
298:
299: \bibitem{grb:04} A. Grbic, and G. V. Eleftheriades, Phys. Rev. Lett. \textbf{%
300: 92}, 117403 (2004).
301:
302: \bibitem{fan:05} N. Fang, H. Lee, C. Sun, and X. Zhang, Science \textbf{308}%
303: , 534 (2005).
304:
305: \bibitem{ves:68} V. G. Veselago, Sov. Phys. Usp. \textbf{10}, 509 (1968).
306:
307: \bibitem{pen:06a} J. B. Pendry and D. R. Smith, Sci. Am. \textbf{295}, 60
308: (2006).
309:
310: \bibitem{pen:06} J. B. Pendry, D. Schurig, D. R. Smith, Science \textbf{312}%
311: , 1780 (2006).
312:
313: \bibitem{sch:06} D. Schurig, J. J. Mock, B. J. Justice, S. A. Cummer, J. B.
314: Pendry, A. F. Starr, D. R. Smith, Science \textbf{314}, 977 (2006).
315:
316: \bibitem{jac:99} J. D. Jackson, \textit{Classical Electrodynamics, }3rd ed.
317: (Wiley, New York, 1999), p. 302.
318:
319: \bibitem{smi:03} D. R. Smith and D. Schurig, Phys. Rev. Lett. \textbf{90},
320: 77405 (2003).
321:
322: \bibitem{ram:03} S. A. Ramakrishna, J. B. Pendry, M. C. K. Wiltshire, and W.
323: J. Stewart, J. Mod. Opt. \textbf{50}, 1419 (2003).
324:
325: \bibitem{bel:06} P. A. Belov and Y. Hao, Phys. Rev. B \textbf{73}, 113110
326: (2006).
327:
328: \bibitem{sal:06} A. Salandrino and N. Engheta, Phys. Rev. B \textbf{74},
329: 075103 (2006).
330:
331: \bibitem{liu:07} Z. W. Liu, H. Lee, Y. Xiong, C. Sun, and X. Zhang, Science
332: \textbf{315}, 1686 (2007).
333:
334: \bibitem{ber:78} D. Bergman, Phys. Lett. \textbf{43}, 377 (1978).
335:
336: \bibitem{ash:76} N. W. Ashcroft and N. D. Mermin, \textit{Solid State Physics%
337: } (Saunders, New York, 1976), p. 156.
338:
339: \bibitem{taf:95} A. Taflove, \textit{Computational Electrodynamics: The
340: Finite-Difference Time-Domain Method} (Artech House, Boston, 1995).
341:
342: \bibitem{ber:94} J. P. Berenger, J. Comput. Phys. \textbf{114}, 185 (1994).
343:
344: \bibitem{joh:72} P. B. Johnson and R. W. Christy, Phys. Rev. B \textbf{6},
345: 4370 (1972).
346: \end{thebibliography}
347:
348: \end{document}
349: