1: %\documentclass[twocolumn,showpacs,aps,amsmath,amssymb,pre,preprintnumbers]{revtex4}
2: \documentclass[preprint,showpacs,aps,amsmath,amssymb,pre,preprintnumbers]{revtex4}
3:
4: % Corrections: (1) English, (2) adding the reference to the M Sc thesis. May 15, 2004.
5:
6: \newcommand{\BE}{\begin{equation}}
7: \newcommand{\EE}{\end{equation}}
8: \newcommand{\BA}{\begin{eqnarray}}
9: \newcommand{\EA}{\end{eqnarray}}
10: \usepackage{graphicx}% Include figure files
11: \usepackage{dcolumn}% Align table columns on decimal point
12: \usepackage{bm}% bold math
13: \input epsf.tex
14: \preprint{To appear in Chinese J. Physics
15: (http://psroc.phys.ntu.edu.tw/cjp/index.php)}% cond-mat/0408477}
16:
17: %\setlength{\textheight}{9.5in} \setlength{\textwidth}{7in}
18: %\setlength{\oddsidemargin}{-1cm} \setlength{\evensidemargin}{-2cm}
19: %\setlength{\topmargin}{-0.8in}
20:
21: \begin{document}
22:
23: \title{About negative refraction and left handed materials}
24:
25: \author{Zhen Ye}\email{zhen@phy.ncu.edu.tw}
26: \affiliation{Wave Phenomena Laboratory, Department of Physics,
27: National Central University, Chungli, Taiwan 32054}
28:
29: \date{May 4, 2004}
30:
31: \begin{abstract}
32:
33: Here I present an overview of recent studies of the phenomenon of
34: negative refraction and left-handed materials. I will discuss some
35: basic questions pertinent to the problem. It is pointed out that
36: the current claims of negative refraction and left handed
37: materials are not conclusive. To support our consideration, I will
38: start with the fundamental physics.
39:
40: \end{abstract}
41: \pacs{78.20.Ci, 42.30.Wb, 73.20.Mf, 78.66.Bz} \maketitle
42:
43: \section{Prelude}
44:
45: Waves surround us. Direct human communication is mainly conveyed
46: by acoustic waves, and is enriched by gestures which are passed
47: into our eyes through optical waves. Indirect human conversation
48: is transmitted via electromagnetic (EM) waves. Nowadays electronic
49: waves are also everywhere in our daily experiences such as audio
50: \& video systems, computers, and Nintendo games. Among many
51: interesting properties pertinent to waves, refraction is probably
52: the most common phenomenon in our daily life.
53:
54: In simple terms, refraction refers to waves bending in passing
55: obliquely from one medium to another. It is a common observation
56: that a beam of light bends when it enters glass or water at a
57: tilted angle. Rainbow which must be a phenomenon known to every
58: one who could see is another example of wave refraction. More
59: specifically, refraction occurs when waves pass across the
60: interface between two different effectively uniform media. Briefly
61: speaking, the effectiveness refers to that inside the medium a
62: phase vector ($\vec{k}$) can be effectively defined for a given
63: frequency. In an isotropic medium, for example, a plane wave can
64: be supported and is described in the form like
65: $A\exp(i\vec{k}\cdot\vec{r}- i\omega t)$. Therefore to study the
66: phenomenon of refraction, it is important to verify whether the
67: media are effective or not.
68:
69: In the world with which we are so far most familiar, EM waves or
70: lights propagate according to the rule of right hand. That is, the
71: electric and magnetic vector fields $\vec{E}$ and $\vec{H}$, and
72: the phase vector $\vec{k}$ form a righthanded coordinate system.
73: In other words, the direction $\vec{k}$ is along the direction of
74: $\vec{E}\times\vec{H}$. Materials in which $\vec{k}$ and
75: $\vec{E}\times\vec{H}$ are in the same direction are called
76: isotropic media. The direction of $\vec{E}\times\vec{H}$ is called
77: the ray or beam direction.
78:
79: In nature, there is another class of materials. In these
80: materials, although the direction of $\vec{k}$ may not be exactly
81: along the direction of $\vec{E}\times\vec{H}$, the angle between
82: the two is smaller than 90 degree, in the mathematic terms,
83: $\vec{k}\cdot(\vec{E}\times\vec{H}) > 0$. This class of materials
84: is called optically anisotropic media. Calcite crystals and
85: YVO$_4$ bi-crystals are two examples of anisotropic media.
86:
87: Wave refraction can appear dramatically different when crossing
88: the interface between either isotropic or anisotropic media. For
89: example, when an optical wave crosses an interface between two
90: isotropic media, the refracted wave will always bend towards the
91: other side of the incidence with regard to the normal of the
92: interface. The refraction and incident angles obey Snell's
93: law\cite{Gen}. When the wave passes an interface between an
94: isotropic medium and an anisotropic medium, however, a phenomenon
95: called amphoteric refraction can occur. That is, the refracted ray
96: may bend either towards the same side as the incidence or the
97: other side of the incidence, depending on the incident angle and
98: the orientation of the anisotropic medium. In this case, Snell's
99: law fails\cite{Yariv}. For a uniaxial crystal, for instance, there
100: are two eigen modes of plane waves: the ordinary and
101: extra-ordinary modes\cite{Yariv}. The dispersion relation of the
102: ordinary mode is isotropic and this mode of light behaves as the
103: light beam in an ordinary material, while the equal frequency
104: surface or normal surface\cite{Yariv} of the extra-ordinary mode
105: is ellipsoidal with the optic axis being the axis of revolution.
106: The allowed wave vectors are those vectors pointing from the
107: center of the ellipsoid to its surface. For this mode, the energy
108: velocity is shown to be equal to the group velocity determined by
109: $\vec{v}_g = \nabla_{\vec{k}}\omega(\vec{k})$\cite{Yariv}.
110:
111: Figure~\ref{fig1} is an illustration of refraction across a
112: boundary between two isotropic media whose refraction indices are
113: denoted as $n_1$ and $n_2$ respectively. In this case, the ray
114: direction and the phase vector's direction are the same. The
115: boundary condition $\vec{k}_1\cdot\vec{r}_\parallel =
116: \vec{k}_2\cdot\vec{r}_\parallel$ leads to Snell's law \BE
117: n_1\sin\theta_i = n_2\sin\theta_r,\label{eq:snell}\EE where
118: $\theta_i$ and $\theta_r$ denote the incidence and refraction
119: angles respectively.
120:
121: \begin{figure}[hbt]
122: \vspace{10pt} \epsfxsize=2in\epsffile{fig1.eps}\smallskip
123: \caption{Illustration of refraction at the interface between two
124: isotropic media.} \label{fig1}
125: \end{figure}
126:
127: Figure~\ref{fig2} depicts the amphoteric refraction of the
128: extraordinary light at a flat interface between an isotropic
129: medium and an anisotropic medium. The boundary condition requires
130: that the phase vectors $\vec{k}$ be continuous, when projected
131: onto the interface. The ray direction, denoted by the time
132: averaged Poynting vector $\vec{S} =
133: \frac{1}{2}\mbox{Re}[\vec{E}\times\vec{H}^\star]$, is normal to
134: the equal frequency surface. Here the refracted ray is bent
135: towards the same side as the incident ray in (a) while the ray is
136: tilted to the other side in (b). Such an amphoteric refraction
137: phenomenon can be readily observed at the interface between air
138: and calcite crystals\cite{Yau}. In this case, Snell's law in
139: Eq.~(\ref{eq:snell}) fails. A modified version of the angle
140: dependence should be introduced\cite{Yau}.
141:
142: \begin{figure}[hbt]
143: \vspace{10pt} \epsfxsize=3.35in\epsffile{fig2.eps}\smallskip
144: \caption{Illustration of amphoteric refraction at the interface
145: between an isotropic medium and an anisotropic medium: (a)
146: negatively appearing refraction; (b) positive refraction}
147: \label{fig2}
148: \end{figure}
149:
150: \section{Introduction}
151:
152: Equipped with the above textbook knowledge, we are ready to
153: discuss the much debated issue of Left Handed Materials (LHMs) and
154: the associated phenomenon of negative refraction (NR).
155:
156: The concept of LHMs was introduced by Veselago many years ago
157: \cite{Ves}. What is Left Handed Material? According to Veselago
158: \cite{Ves}, a left handed material refers to electrodynamics of
159: substance with simultaneously negative values of permittivity
160: $\epsilon$ and permeability $\mu$. An immediate deduction from
161: this perception is that the material should have a negative
162: refraction index, as $n = \sqrt{\epsilon}\sqrt{\mu} = -
163: \sqrt{|\epsilon\mu|}$. Another important outcome is that the
164: energy ray $\vec{S}$ and the phase vector $\vec{k}$ will no longer
165: point to the same direction. Instead, they will be opposite to
166: each other in an isotropic LHM or make an angle bigger than 90
167: degree in an anisotropic LHM, i.~e. $\vec{k}\cdot \vec{S} < 0$.
168:
169: A negative refractive index would lead to a few peculiar
170: phenomena. Taking the propagation of EM waves from a normal
171: material through an isotropic LHM as the example, some of the
172: expectations are described in Fig.~\ref{fig3}. Inside the LHM, the
173: phase vector points (Black arrows) to the opposite direction of
174: the energy rays, denoted by the red arrows. Fig.~\ref{fig3}(a)
175: illustrates the negative refraction. Here, Snell's law applies
176: as\BE n_1\sin\theta_i = (-|n_2|)(-|\sin\theta_r|).\EE
177:
178: The second prediction is the imaging by a slab of LHM as shown in
179: Fig.~\ref{fig3}(b). For a slab of normal refractive materials,
180: isotropic or anisotropic, no image can be made across the slab.
181: There will be no image inside the slab either. Here,
182: Fig.~\ref{fig3}(b) shows that two images can be made across the
183: slab of LHM when a wave is transmitted from a point source.
184:
185: \begin{figure}[hbt]
186: \vspace{10pt} \epsfxsize=3.25in\epsffile{fig3.eps}\smallskip
187: \caption{Illustration of a few expected phenomena for EM wave
188: propagation through a LHM: (a) negative refraction; (b) Imaging by
189: a flat slab of LHM; (c) negative refraction by a prism of made of
190: LHM; (c) negatively shifted outgoing waves. Here the red arrows
191: denote the energy ray inside the LHM. Here we have ignored the
192: high order internal reflections.} \label{fig3}
193: \end{figure}
194:
195: The third possibility is the negative refraction by a prism of
196: LHMs. This is described in Fig.~\ref{fig3}(c). A plane wave is
197: incident to the horizontal side of the prism. At the long side,
198: the incident angle is determined as the prism angle. From the
199: refraction angle $\theta_r$, one can deduced the refractive index
200: using Snell's law. If it is tilted towards the left side, the
201: refractive index is negative. Otherwise, the refractive index is
202: positive.
203:
204: When a plane wave is incident on to a flat slab of LHMs, the
205: outgoing wave will be shifted towards the left side, as shown in
206: Fig.~\ref{fig3}(d). The negative refractive index can thus be
207: deduced as \BE n_2 =
208: -\frac{n_1\sin\theta_i}{\sin(\tan^{-1}(D/L))}. \label{eq:flat}\EE
209:
210: \begin{figure}[hbt]
211: \vspace{10pt} \epsfxsize=3.25in\epsffile{fig4.eps}\smallskip
212: \caption{Illustration of amphoteric refraction of the
213: extra-ordinary waves at the interface between an isotropic normal
214: medium and an anisotropic uniaxial LHM medium: (a) positive
215: refraction; (b) negative refraction. The black and red arrows
216: denote the phase vectors and energy rays repectively. In isotropic
217: normal media, the energy ray follows the direction of the phase
218: vector.} \label{fig4}
219: \end{figure}
220:
221: The four deductions described in Fig.~\ref{fig3} have actually
222: formed the base line in the current searching for LHMs and
223: negative refraction. A number of notes must be made.
224: \begin{enumerate}
225:
226: \item An important component underlying these scenarios is that
227: the medium can be regarded as an effective medium, in which a
228: refractive index can be defined, and all the wave propagation will
229: be fully describable by the Maxwell equations based upon this
230: index.
231:
232: \item Another essential point is that the LHM is isotropic, to
233: warrant the four deductions. For an anisotropic LHM material, the
234: amphoteric refraction can also appear, as shown in
235: Fig.~\ref{fig4}. In this case, the only way to differentiate the
236: amphoteric refraction by a normal anisotropic medium from the
237: amphoteric refraction by an anisotropic LHM medium is to look at
238: the phase information. In addition, Snell's law fails for the
239: anisotropic case; therefore the refractive index cannot be readily
240: obtained in this case.
241:
242: \item The negative refraction of LHMs in Fig.~\ref{fig3} must be
243: differentiated from the negatively appearing refraction due to
244: anisotropy in Fig.~\ref{fig2}(a). The mechanisms behind the two
245: similar phenomena are completely different. As shown, the phase
246: information in the two scenarios is totally different. The
247: negative refraction of LHMs would be revolutionary not only with
248: respect to technology, but also to fundamental physics such as
249: quantum electrodynamic processes. To avoid confusion, we call the
250: negatively appearing refraction due to anisotropy or any non-LHM
251: related effect {\bf quasi-negative refraction}, while reserving
252: the term {\bf negative refraction} for LHMs. Negative refraction
253: without negative index belongs to the former category. The two
254: categories are completely different in the mechanism.
255:
256: \end{enumerate}
257:
258: The proposal of LHMs by Veselago did not attract much attention
259: until 2000 when Pendry further pointed out that negative
260: refraction makes a perfect lens\cite{Pendry}. Ever since then, the
261: research on such a perfect lens and LHM has been skyrocketing in
262: the midst of much debate. A great body of literature has been and
263: continues to be generated.
264:
265: In spite of a few challenges with regard to the concept of LHM or
266: relevant negative refraction effects \cite{Garcia1,PE,PE1,YePRE},
267: the mainstream consensus has been that some indications of
268: negative refraction effects are affirmative\cite{Pendry2003}:
269: positively negative. Upon inspection, however, we believe that
270: previous evidence for negative refraction effects may not be
271: sufficient. The negatively appearing refraction may not be caused
272: by negative refraction. In fact, some of the phenomena perhaps
273: should not have been considered in the context of refraction.
274: Instead, they may be better characterized such as diffraction,
275: deflection or the anisotropic scattering phenomenon. In the
276: present paper, we wish to give a detailed elaboration of this
277: point of view. Some of present results can be referred to our
278: unpublished preprints available from the internet.
279:
280: \section{A brief review of the research on negative refraction}
281:
282: There are two approaches towards the issue of negative refraction
283: and LHMs. One is to explore all possible properties or
284: applications of LHMs. This approach is meaningful if and only if
285: LHMs indeed exist. The second approach, more difficult but
286: important and fundamental, is to search for or fabricate LHMs. In
287: this paper, I will only consider the second approach.
288:
289: So far, there is only one experimental report about naturally
290: existing materials for which the negatively appearing refraction
291: has been vaguely related to the negative refraction\cite{Zhang}.
292: As discussed later\cite{Yau,Comm}, the negatively appearing
293: refraction is due to the anisotropy of the materials, and is the
294: quasi-negative refraction phenomenon as illustrated in
295: Fig.~\ref{fig2}. It has been pointed out above, it is important to
296: differentiate the quasi-negative and negative refractions. Failing
297: to recognize the difference could cause significant
298: confusion\cite{Report1,Report2}.
299:
300: In searing for LHMs, two types of artificial materials have been
301: proposed so far. One type involves metallic wires and split ring
302: resonators, first made and experimentally tested by Shelby {\it et
303: al.}\cite{Smith}, later further supported by other groups (e.~g.
304: Refs.~\cite{Houck,Parazz}). The second type is the two dimensional
305: photonic crystals made by arrays of parallel cylinders, initiated
306: theoretically by Notomi\cite{Notomi} and Luo {\it et
307: al.}\cite{Luo}, later experimentally tested by Cubukcu, {\it et
308: al.} and Parimi {\it et al.}\cite{Cubukcu,Sridhar,Sridhar2}. Not
309: withstanding or ignoring other investigations, these explorations
310: are the basis for later experimental or theoretical adventures in
311: supporting the negative refraction or LHMs.
312:
313: All the explorations on LHMs and negative refraction are based on
314: at least one of the four scenarios shown in Fig.~\ref{fig3}. For
315: example, we can list as follows: (a) Ref.~\cite{Notomi}; (b)
316: Refs.~\cite{Houck,Luo,Sridhar}; (c)
317: Refs.~\cite{Smith,Houck,Parazz}; (d) Ref.~\cite{Cubukcu}. In
318: particular, the negative refraction index has been obtained for
319: the resonating wire and ring systems in
320: Refs.~\cite{Smith,Houck,Parazz} and for photonic crystals in
321: Ref.~\cite{Cubukcu,Sridhar2}.
322:
323: \section{Discussion}
324:
325: \subsection{Foundation of consideration}
326:
327: The question we wish to discuss is whether the current evidence
328: for LHMs and the related negative refraction is sufficient or
329: conclusive. Moreover, if further confirmation is needed, what
330: could be done next. Our consideration will be based upon the
331: following logic reasoning.
332: \begin{enumerate} \item Point I: If a phenomenon has another
333: possible explanation, then it is insufficient to attribute the
334: phenomenon to a particular mechanism. In other words, if C appears
335: to be a common phenomenon which could be due to A and B. Then C
336: cannot be regarded as the unique consequence of either A or B
337: alone. It is also insufficient to attribute C to either A or B.
338: Such a degeneracy has to be verified by finer scrutiny. As
339: mentioned earlier, measurements at various angles is to isolate
340: isotropy effects, and the phase information is important to
341: differentiating the quasi-negative refraction due to anisotropy
342: and negative refraction by anisotropic LHMs. Like in quantum
343: mechanics, the angular momentum measurement is important to
344: differentiate different states at the same energy level when the
345: conservation law of angular momentum holds. \item Point II: The
346: evidence should not have any self-conflicting point. \item Point
347: III: The observation should not be accidental. Further
348: confirmation should be made to check for robustness, unless the
349: evidence is sufficient. \item Point IV: The theoretical support
350: should be independent. That is, the theoretical analysis should be
351: {\it ab initio}, i.~e. the simulation should not use parameters
352: from the experiment, unless there are new predictions for
353: experiments to verify. \item Point V: Experimental verifications
354: with similar setups and conditions may not be considered as
355: completely independent.
356:
357: \end{enumerate}
358:
359: These points are important. As an illustration, let us consider
360: the example in Fig.~\ref{fig2}. If only measuring the energy rays
361: (red arrows) in case (a), then applying Snell's law to the rays,
362: one may claim that a negative refraction has been observed and a
363: negative refractive index can be deduced. This declaration is
364: inappropriate.
365:
366: First, the claim has to rely on the following assumptions: (1) the
367: medium is isotropic and can be described as an effective medium;
368: so Snell's law is applicable; (2) based upon (1), it is expected
369: that negative refraction can also be observed at other incident
370: angles even without confirmation. Without verifying the validity
371: of these two assumptions, one will be unable to exclude the
372: possibility that the observation is accidental (Point III).
373:
374: Second, if simply taking the measured data from Fig.~\ref{fig2}(a)
375: and taking the negative refractive index from applying Snell's
376: law, one would be able to reproduce the ray direction measured and
377: may claim as a theoretical support. Clearly, such a theoretical
378: support is not independent (Point IV). In addition, the support
379: may work for one particular measurement, but fail in other
380: extended measurements. For example, an isotropic negative
381: refractive index would fail to explain the observation in
382: Fig.~\ref{fig2}(b). Now if one comes back to show that the
383: refraction is anisotropic, then this will be in conflict with the
384: presumption (Point II).
385:
386: Third, the negatively appearing refraction in Fig.~\ref{fig2}(a)
387: can be explained by the medium anisotropy. Therefore it is
388: insufficient to declare that the observation is negative
389: refraction, with the implication of finding a LHM (Point I).
390:
391: Fourth, different groups may measure the energy rays with
392: different locations of the detector but with the same setup and
393: condition, and then claim to reproduce each other's results. This
394: type of supports is clearly not independent (Point V).
395:
396: These considerations will guide our following discussion. We must
397: stress again that we have no reason to criticize the experimental
398: evidence and theoretical analysis, but we wish to consider whether
399: they are sufficient or conclusive. If not, what can be done to
400: improve.
401:
402: \subsection{Analysis of the evidence and methodology}
403:
404: \subsubsection{General discussion}
405:
406: In all the experiments that have been performed so far, only the
407: wave intensity field, which is proportional to the square modulus
408: of the EM field, has been measured. There is no report of the
409: experimental or theoretical analysis of the phase information of
410: the systems which are claimed to be either LHMs or negatively
411: refracting. We note that there are theoretical investigations of
412: the phase behavior of LHMs based on the {\it a priori} assumption
413: that the systems are LHMs; we will not consider this type of
414: investigations.
415:
416: Probing the refraction phenomenon and obtaining refractive index,
417: without the phase information, will have to rely on the assumption
418: that the medium must be effective and isotropic. So far, there is
419: no verification on the isotropy and effectiveness. A few possible
420: ways to check for the isotropy and effectiveness. The isotropy can
421: be checked by rotating a sample of the medium. If a medium is
422: effective and isotropy, the wave propagation inside the medium
423: will not be sensitive to the fine structures of the medium, and
424: therefore it will not be changed when the structures are uniformly
425: disordered. So far, no experiment has been done along this line.
426: As pointed above, it is improper to use Snell's law to energy ray
427: paths without confirming the isotropy and effectiveness.
428:
429: \subsubsection{Negative refraction by composite materials}
430:
431: The composite materials used in Ref.~\cite{Smith,Houck,Parazz} are
432: made of periodically arranged metallic wires and and split ring
433: resonators. The components form a two dimensional square
434: crystal-lattice. The EM waves are transmitted along the direction
435: perpendicular to the axis of the wires. The measurements have
436: adopted the prism scenario described in Fig.~\ref{fig3}(c). The
437: energy rays are measured and used in Snell's law to obtain the
438: negative refraction index.
439:
440: While some questions have already raised in Ref.~\cite{PE,PE1}, a
441: number of other questions can be raised with this type of
442: measurements. First, there is no verification for the isotropy of
443: the media, which could be readily verified by the flat slab
444: transmission scenario in Fig.~\ref{fig3}(d). Due to their periodic
445: structured, the materials are unlikely isotropic. If so, it would
446: not have been proper to invoke Snell's law for the energy rays
447: (Point II). Second, there is no information about either the wave
448: propagation inside the media nor about the phase. One possible
449: solution is to correlate the phase of the outgoing waves with that
450: of the incident waves. Third, since the composite materials are
451: made of regularly arranged split ring resonators, possible band
452: structure effects need to be considered. The essence is to show
453: that the structured materials can be indeed regarded as an
454: effective medium in which a plane wave can be established. This
455: question has already been raised by Pokrovsky and
456: Efros\cite{PE,PE1}. In addition, the interpretation of the
457: observation in Ref.~\cite{Smith} has also been questioned for
458: possible effects due to absorption\cite{Garcia1}.
459:
460: In Ref.~\cite{Houck}, two prisms of the composite materials are
461: used. The energy rays are measure and used to deduce the
462: refraction index. The deviation between the two cases amounts to
463: at least 40 percent. Reference \cite{Parazz} duplicates the
464: experiment in Ref.~\cite{Smith} (Point V). The numerical
465: simulation is not independent (Point IV). As pointed out in
466: Ref.~\cite{PE1}, a correct computer simulation starting with the
467: microscopic Maxwell's equation would reproduce the experiments,
468: but whether the observations can be explained in terms of a
469: macroscopic Maxwell's equation with both negative permittivity and
470: permeability remains an open question. Such an {\it ab initio}
471: simulation is yet to come. In addition, as Pokrovsky and Efros
472: pointed out\cite{PE1}, there are no doubts on the the results, but
473: there is a need for solid confirmation of the data interpretation.
474:
475: \subsubsection{Negative refraction by photonic crystals}
476:
477: The first proposals of negative refraction by photonic crystals
478: were from Notomi\cite{Notomi} and Luo {\it et al.}\cite{Luo},
479: later experimentally tested by Cubukcu, {\it et al.} and Parimi
480: {\it et al.}\cite{Cubukcu,Sridhar,Sridhar2}.
481:
482: Most of photonic crystals studied so far are two dimensional
483: arrays of dielectric cylinders. An advantage of this type of
484: systems is that it can be solved {\it exactly} by the multiple
485: scattering theory (MST), first formulated by
486: Twersky\cite{Twersky}, later reformulated for EM
487: systems\cite{JOSA,Bikash}, acoustic systems\cite{ChenPRL}, and
488: water waves\cite{Yue}. Therefore we can have a rigorous
489: examination or new predictions of this type of photonic crystals.
490: We will adopt MST.
491:
492: {\it Imaging inside a photonic crystal.} By the FDTD simulation,
493: Notomi presented an image focusing inside a photonic crystal made
494: of drilled holes in a uniform dielectric material. An focused
495: imaging is presented in Fig.~13 of Ref.~\cite{Notomi}. We have
496: tried to reproduce the results, but failed, partially because the
497: parameters have not been clearly given in Ref.~\cite{Notomi}. In
498: the simulation, however, we found that the snapshots of the
499: imaging field depend significantly on the simulation time. When
500: the stable stage has not been reached, some focused images like in
501: \cite{Notomi} could be exhibited\cite{Tang}.
502:
503: \begin{figure}[hbt]
504: \vspace{10pt} \epsfxsize=3.25in\epsffile{fig5.eps}\smallskip
505: \caption{(a1) and (a2): The imaging fields for a slab of photonic
506: crystal structure. The slab measures as
507: 10$\sqrt{2}\times40\sqrt{2}$. (b1) and (b2): The imaging fields
508: for a transmitting source located inside a square array of
509: cylinders. The square measures $14\sqrt{2}\times 14\sqrt{2}$.}
510: \label{fig5}
511: \end{figure}
512:
513: {\it Flat lens imaging - theoretical.} The flat lens imaging
514: illustrated by Fig.~\ref{fig3}(b) was first studied in
515: Ref.~\cite{Luo}. The systems considered are two types of photonic
516: crystals. One is made of square arrays of dielectric cylinders in
517: air. The other is made of square arrays drilled holes in a
518: dielectric medium. Based upon the equal frequency band structure,
519: the authors show there is a range of frequencies in which an all
520: angle negative refraction can be archived in this scenario. We
521: have done two verifications. First, we have reproduced nicely the
522: observation of Ref.~\cite{Luo}. The results are shown in Fig.~1(c)
523: of Ref.~\cite{Kuo2}. We indeed observe a focused image across the
524: flat slab of photonic crystals.
525:
526: If such an imaging property is due to negative refraction, we
527: would then have two expectations (Point III): one is to find a
528: focused image inside the slab as shown in Fig.~\ref{fig3}(b); the
529: other is to see the negative refraction effect in the scenario
530: shown in Fig.~\ref{fig3}(d).
531:
532: For the first expectation, the inside focusing is hard to see in
533: the arrays of Ref.~\cite{Luo}, as their photonic crystal slab is
534: too thin. We have thus enlarged the slab and plot both the
535: amplitude fields and the intensity fields inside the slab. There
536: is no focused image inside the slab. Instead, we observe that the
537: energy transmission is guided through the slab, making the image.
538: This has be further verified by putting an EM wave source inside a
539: square shaped array of cylinders. The two sets of simulation
540: results, taken from Ref.~\cite{Kuo2}, are shown in
541: Fig.~\ref{fig5}. All the parameters and lattice arrangements are
542: taken from Ref.~\cite{Luo}; all lengths are scaled by the lattice
543: constant and the frequency is scaled by the radius of the
544: cylinders. Here we see clearly the image focused on the other side
545: of the slab in (a) and the four images made at the four exiting
546: points are due to the guided or collimated wave transmission. It
547: appears that there are favorable directions for waves to
548: transport, as shown by (b2). We found that such collimated or
549: guided transmission is caused by the presence of partial bandgaps.
550: As can be seen from Fig.~3 of Ref.~\cite{Luo}, the frequency
551: chosen lies within a partial bandgap in which the waves are
552: forbidden to transmit along the direction of $\Gamma-X$ in the
553: square lattice. Due to this partial bandgap, waves seem to avoid
554: the $\Gamma-X$ direction as much as possible, and thus appear to
555: be collimated along the $\Gamma M$ direction. The energy paths
556: shown in Fig.~\ref{fig5} are actually along the $\Gamma-M$
557: direction.
558:
559: We have also checked the second expectation. The results are
560: presented in Ref.~\cite{Chien}. We found that the wave
561: transmission always tends to be guided or collimated into the
562: $\Gamma M$ direction, independent of the incident angles or the
563: orientation of the $\Gamma-M$ direction. This phenomenon cannot be
564: classified as refraction either in the context of isotropy or
565: anisotropy.
566:
567: The results from Refs.~\cite{Kuo2,Chien} led us to propose that
568: partial bandgaps can serve as a guiding tunnel for the wave
569: transport, allowing for new applications of photonic
570: crystals\cite{Chen}. Such a consideration has also been extended
571: to acoustic systems\cite{Chen5}.
572:
573: The flat slab imaging discussed in Ref.~\cite{Luo} is not caused
574: by negative refraction, rather it is caused by directed
575: diffraction. The misinterpretation in Ref.~\cite{Luo} has also
576: been pointed out by Li {\it et al.}\cite{Li}. At least, the
577: explanation in Ref.~\cite{Luo} is unable to exclude other possible
578: interpretations (Point I).
579:
580: One may ask for the reason for the discrepancies between the
581: present results and the analysis in Ref.~\cite{Luo}. Here we would
582: like to share our thoughts. The problem with the conjectured flat
583: slab imaging may lie in the approach to the energy flow inside the
584: slab. The usual approach mainly relies on the curvatures of
585: frequency bands to infer the energy flow.
586:
587: As documented in Ref.~\cite{Yariv}, an energy velocity is defined
588: as $\vec{v}_e = \frac{\frac{1}{V}\int \vec{J}_{\vec{K}}
589: d^3{r}}{\frac{1}{V}\int U_{\vec{K}} d^3{r}},$ where
590: $\vec{J}_{\vec{K}}$ and $U_{\vec{K}}$ are the energy flux and
591: energy density of the eigenmodes, and the integration is performed
592: in a unit cell, representing a spatially averaged value. It can be
593: shown that thus defined energy velocity equals the group velocity
594: obtained as $\vec{v}_g = \nabla_{\vec{K}}\omega(\vec{K}).$
595: Therefore it is common to calculate the group velocity to infer
596: the energy velocity and subsequently the energy flows or
597: refraction of waves. Whether the net current flow through a unit
598: cell really follows the direction of $\vec{v}_e$ remains unclear.
599: We note here that the average flux through a surface may be
600: defined as $\langle\vec{J}\rangle = \frac{\hat{n}}{S}\int
601: d\vec{S}\cdot \vec{J}$, where $\hat{n}$ is the unit normal vector
602: of the surface $S$. Clearly, the volume averaged current within a
603: unit cell does not necessarily correspond to actual energy
604: flows\cite{GV,PLA}. General speaking, the periodically structured
605: materials may not be regarded as an effective medium.
606:
607: The curvature approach fails in the current partial bandgap case.
608: For example, in the present case, the curvature method would lead
609: to an almost 80 degree of energy deflection with reference to the
610: $\Gamma X$ direction when the incidence is at about 25 degree; The
611: rigorous computational results show that the energy flow is nearly
612: in the $\Gamma M$ direction\cite{Chien}.
613:
614: \begin{figure}[hbt]
615: \vspace{10pt} \epsfxsize=3.25in\epsffile{fig6.eps}\smallskip
616: \caption{The image of the intensity-fields for flat slabs. (a) and
617: (b) refer to two arrangements. In (a), all the arrangements
618: including the source, the field resolution, and the cylinders are
619: identical to the experiment\cite{Sridhar}. (b) is the same as (a)
620: except that the source is move upward by a half lattice constant.
621: The scales for the fields for the left and right sides are shown
622: in the figure.} \label{fig6}
623: \end{figure}
624:
625: {\it Flat lens imaging - experimental.} Later, flat slab imaging
626: has been experimentally observed by Parimi et al.\cite{Sridhar}.
627: The authors reported that an image can be formed cross a flat slab
628: of photonic crystals made of dielectric cylinders. The appearance
629: of the image is regarded as the occurrence of negative refraction
630: and has thus been related to the negative refraction of LHMs. We
631: found that their interpretation needs to be reconsidered. The
632: focused image is accidental (Point III). The imaging pattern will
633: completely change as the source is moved only by a half lattice
634: constant. From the transmission calculation, we found that the
635: image is due to the anisotropic scattering of the array of
636: cylinders. Scattering by a group of materials certainly will yield
637: bright-dark patterns, simply due to multiple scattering and wave
638: interference, like in the X-ray imaging. Our rejected comments on
639: the experimental observation are in Ref.~\cite{Kuo7}. Partial
640: results are shown in Fig.~\ref{fig6}. Fig.~\ref{fig6}(a)
641: reproduces remarkably well the experiment. In the case (b), the
642: source is move upward by a half lattice constant. Here we see that
643: the imaging field across the slab changes completely. Again, we do
644: not criticize the experiment, but we think that the interpretation
645: is ambiguous.
646:
647: What is the difference between the flat slab imaging due to
648: partial gaps and the imaging due to anisotropic scattering? The
649: difference is that the imaging by partial gaps is a stable
650: phenomenon, while the imaging due to scattering is unstable. The
651: focusing phenomenon can be even seen by scattering by a single
652: object under certain manipulation, as long as the scattering is
653: anisotropic.
654:
655: {\it Plane wave through a slab of photonic crystals.} Cubukcu {\it
656: et al.} reported the measurement of a beam of EM waves incident on
657: a flat slab of photonic crystals\cite{Cubukcu}, referring to
658: Fig.~\ref{fig3}(d). The authors measured the energy ray directions. Then
659: Equation (\ref{eq:flat}) was used to obtain
660: the refractive index. The author shows that at certain ranges of
661: incident angles, the index is negative. As discussed earlier, Snell's law in
662: the form of Eq.~(\ref{eq:flat}) is only applicable when the medium
663: isotropic. However, the results in Ref.~\cite{Cubukcu} indicate
664: the opposite. This is clearly self-conflicting (Point II). In
665: addition, we tried to reproduce the results in Fig.~2 of
666: Ref.~\cite{Cubukcu}, but failed. We found the results similar to that
667: in Ref.~\cite{Chien}. That is, independent of the incident
668: angles, the waves are bent towards a particular direction, which
669: is across the slab of photonic crystals used in the experiment. A
670: possible reason for the discrepancy may be due to the difference
671: between the FDTD method used in \cite{Cubukcu} and the MST used
672: in our simulation. As mentioned earlier, it takes a considerable long
673: time for FDTD results to reach the stable
674: stage. Without checking the stability, the FDTD results may be
675: unreliable. Another possible cause is the finite size effect.
676: In our simulation, we have excluded any thinkable
677: numerical artifacts, with regards to the mode numbers, finite number of scatterers,
678: boundary effects, and so on.
679: We welcome any request for the
680: numerical codes and our simulation procedures for independent and unbiased verification.
681:
682: {\it Prism scenario.} Lastly but not least, we address the prism
683: scenario. It is obvious that this scenario can only work when the
684: medium is isotropic. Without first verifying that the medium is
685: isotropic, the deduction of the refraction index is not
686: reliable, as the waves may be already bent at the entering interface. As an example, we
687: have simulated two prisms of the photonic crystals made of dielectric cylinders, given in
688: Ref.~\cite{Luo,Kuo2}. The results are presented in
689: Ref.~\cite{Kuo3}. Here we show partial results.
690:
691: We have plotted the transmitted intensity fields. The results are
692: presented in Fig.~\ref{fig7}. Here the incident waves are
693: transmitted vertically from the bottom. The impinging frequency is
694: 0.192$*2\pi c/a$ with $a$ being the radius of the cylinder; the
695: frequency has been scaled to be non-dimensional in the same way as
696: in Ref.~\cite{Luo}. We note that at this frequency, the wave
697: length is about five times of the lattice constant, nearly the
698: same as that used in the experimental of \cite{Houck}. The
699: incidence is along the [$\cos 22.5^o$, $\sin 67.5^o$] direction,
700: that is, the incidence makes an equal angle of 22.5$^o$ with
701: regard to the [10] and [11] directions of the square lattice; the
702: reason why we choose this direction will become clear from later
703: discussions. Moreover, on purpose, the intensity imaging is
704: plotted for the fields inside and outside the prisms on separate
705: graphs which have been scripted by `1' and `2' respectively.
706:
707: \begin{figure}[hbt]
708: \vspace{10pt} \epsfxsize=3.25in\epsffile{fig7.eps}\smallskip
709: \caption{ The images of the transmitted intensity fields. Here,
710: the intensities inside and outside the prism structures are
711: plotted separately, so that the images can be clearly shown with
712: proper scales. The geometric measurements can be inferred from the
713: figure. The tilt angles for the two prisms are 18$^o$ and 26$^o$
714: respectively, and have been labelled in the figures. For cases
715: (a1) and (b1), we observe the apparent `negative refraction' at
716: the angles of 15$\pm 3^o$ and $21\pm 5^o$ respectively. When
717: applying Snell's law, these numbers give rise to the negative
718: refraction indices of $-0.84 \pm 0.17$ and $-0.87\pm 0.21$ for
719: (a1) and (b1) separately. In (a2) and (b2), the intensity fields
720: inside inside the prisms are plotted. Here we clearly see that the
721: transmission has been bent. In the plots, [10] and [11] denote the
722: axial directions of the square lattice of the cylinder arrays,
723: i.~e. $\Gamma-X$ and $\Gamma-M$.} \label{fig7}
724: \end{figure}
725:
726: First, we ignore the transmitted fields inside the prisms. From
727: Fig.~\ref{fig7} (a1) and (b1) we are able to calculate the main
728: paths of the transmitted intensities. The geometries of the
729: transmission are indicated in the diagrams. The tilt angles of the
730: prisms are denoted by $\phi$, whereas the angles made by the
731: outgoing intensities relative to the normals of the titled
732: interfaces are represented by $\theta$. According to the
733: prescription outlined in Ref.~\cite{Smith,Houck}, once $\phi$ and
734: $\theta$ are determined, Snell's law is applied to determine the
735: effective refractive index: $n\sin\phi = \sin\theta$. If the angle
736: $\theta$ is towards the higher side of the prisms with reference
737: to the normal, the angle is considered positive. Otherwise, it is
738: regarded as negative. In light of these considerations, the
739: apparent `negative refractions', similar to the experimental
740: observations\cite{Smith,Houck}, indeed appear and are indicated by
741: the black arrows in the graphs. After invoking Snell's law, the
742: negative refraction results are deduced. From (a1) and (b1), we
743: obtain the negative refractive indices as $-0.84\pm 0.17$ and
744: $-0.87\pm 0.21$ for the two prisms respectively. The inconsistency
745: between the two values is less than 4\%. The overall uncertainty
746: in the present simulation is less than 24\%. Therefore, a
747: consistent negative refractive index may be claimed from the
748: measurements shown in Fig.~\ref{fig7} (a1) and (b1). Furthermore,
749: we found that with certain adjustments such as rotating the arrays
750: or varying the filling factor, the index obtained by Snell's law
751: can be close to the perfect -1.
752:
753: Now we take into consideration the wave propagation inside the
754: prisms. As shown by Fig.~\ref{fig1} (a2) and (b2), the
755: transmission inside the prisms has been bent at the incidence
756: interfaces, referring to the two white arrowed lines in (a2) and
757: (b2). If Snell's law were valid at the outgoing interfaces, it
758: should also be applicable at the incident boundaries. Then with
759: the zero incidence angle and a finite refraction angle, Snell's
760: law would lead to an infinite refractive index for the surrounding
761: medium. Besides, when taking into account the bending inside the
762: prisms, the incident angle at the outgoing or the tilted surface
763: is not $\theta$ any more. Therefore the index value obtained above
764: is incorrect. Our opinion is that we need to differentiate the
765: refraction from diffraction. The phenomenon shown in
766: Fig.~\ref{fig7} should be attributed as occurrence of diffraction
767: effects.
768:
769: The measurements in Ref.~\cite{Sridhar2} have adopted the prism
770: scenario. The photonic crystals are made of cylinders, and made
771: into a prism. Then the energy rays are measured from the incident
772: and outgoing waves, to obtain refractive index by Snell's law.
773: Again, the authors have not verified the isotropy. How waves
774: propagate inside the prism has not been probed. According to the
775: results in Fig.~\ref{fig7}, the experiment is not sufficient to
776: identify the LHM behavior of the photonic crystals.
777:
778: \section{Summary}
779:
780: In summary, we here discuss some fundamental questions about LHMs
781: and the associated negative refraction. It is pointed out that the
782: current experimental evidence is not sufficient to conclude that
783: LHMs have been formulated. Some concerns and considerations are
784: presented to support the analysis. The key is to carry out a
785: comprehensive experimental exploration, with regard to all aspects
786: of LHMs and negative refraction, including such issues as phase
787: information, isotropy, and effectiveness etc. The phase
788: information will be essential to differentiate the negative
789: refraction from the quasi-negative refraction.
790:
791: Before talking about either the positive or negative refraction, a
792: prior task is to find out whether it is the refraction phenomenon
793: that occurs when waves pass across the interface between two
794: media. It is insufficient to just look at the deflection of the
795: energy intensity path to discern the refraction phenomenon.
796: Otherwise, confusions can accumulate\cite{Report3}. In sonic
797: crystals, it can be shown that negative refraction can never occur
798: in the sense of LHMs. We think that the report in
799: Ref.~\cite{Report3} misleading. Structured materials can certainly
800: manipulate wave propagation in various ways. However exciting, it
801: is crucial to confirm first the manipulation is due to the
802: refraction mechanism before classifying as negative or positive
803: refraction, and claiming the finding or fabricating of LHMs.
804:
805: The essence of the present work is not to discard the previous
806: efforts, rather it is hoped to help clarify some of possible
807: uncertainties in the previous data interpretation or theoretical
808: analysis. Only through the detailed examination, debate or
809: introspection, the understanding of the problem can be deepened.
810: Given the situation that the issue of LHMs has ignited such a huge
811: response, a careful reconsideration of the problem is always
812: something important.
813:
814: {\bf Acknowldegments} I am grateful to Ting-Kuo Lee, Ding-Ping
815: Tsai, Han-Fei Yau, Yong Zhang, Ben-Yuan Gu, Zhi-Yuan Li,
816: Chao-Hsien Kuo, Liang-Shan Chen, M. Nieto-Vesperinas, Dezhang Chu
817: and many others for constructive correspondences and criticisms.
818: The help from C.-H. Kuo is particularly thanked. The work received
819: support from NSC, and the College of Science at NCU.
820:
821: \begin{thebibliography}{99}
822:
823: \bibitem{Gen} P. G. Hewitt, {\it Conceptual Physics},
824: (Addison-Wesley, New York, 2002).
825:
826: \bibitem{Yariv} A. Yariv and P. Yeh, {\it Optical Waves in Crystals}, (John Wiley \&
827: Sons, Inc., New York, 1984).
828:
829: \bibitem{Yau} B. Ke, M. Sc. thesis for NCU (2003); available at
830: http://thesis.lib.ncu.edu.tw.
831:
832: \bibitem{Ves} V. G. Veselago, Sov. Phys. Usp. {\bf 10}, 509
833: (1968).
834:
835: \bibitem{Pendry} J. B. Pendry, Phys. Rev. Lett. {\bf 85}, 3966
836: (2000).
837:
838: \bibitem{Garcia1} N. Garcia and M. Nieto-Vesperinas, Optics Letts.
839: {\bf 27}, 885 (2002).
840:
841: \bibitem{PE} A. L. Pokrovsky and A. L. Efros, Phys. Rev. Lett.
842: {\bf 89}, 093901 (2002); R. Marqu\'es and D. R. Smith, Phys. Rev.
843: Lett. {\bf 92}, 059401 (2004).
844:
845: \bibitem{PE1} A. L. Pokrovsky and A. L. Efros, Phys. Rev. Lett. {\bf 92},
846: 119401 (2004).
847:
848: \bibitem{YePRE} Z. Ye, Phys. Rev. B {\bf 67}, 193106 (2003)
849:
850: \bibitem{Pendry2003} J. B. Pendry, Nature, {\bf 423}, 22 (2003).
851:
852: \bibitem{Zhang} Y. Zhang, B. Fluegel, and A. Mascarenhas, Phys. Rev. Lett. {\bf 91},
853: 157404 (2003).
854:
855: \bibitem{Comm} Yu.P.Bliokh and J.Felsteiner, physics/0311066.
856:
857: \bibitem{Report1} http://physicsweb.org/article/news/7/10/10.
858:
859: \bibitem{Report2} http://www.aip.org/mgr/png/2003/202.htm.
860:
861: \bibitem{Smith} R. A. Shelby, D. R. Smith, and S. Schultz, Science
862: {\bf 292}, 77 (2001).
863:
864: \bibitem{Houck} A. Houck, J. B. Brock, and I. L. Chuang, Phys. Rev. Lett. {\bf 90},
865: 137401 (2003).
866:
867: \bibitem{Parazz} C. G. Parazzoli, {\it et al}, Phys. Rev. Lett. {\bf 90}, 107401 (2003)
868:
869: \bibitem{Notomi} A. Notomi, Phys. Rev. B {\bf 62}, 10696 (2000).
870:
871: \bibitem{Luo} C. Luo, S. G. Johnson, J. D. Joannopoulos, and J. B. Pendry, Phys.
872: Rev. B {\bf 65}, 201104(R) (2002).
873:
874: \bibitem{Cubukcu} E. Cubukcu, K. Aydin, and E. Ozbay, S. Foteinopolou1 and C.M.
875: Soukoulis, Phys. Rev. Lett. {\bf 91}, 207401 (2003).
876:
877: \bibitem{Sridhar} P. V. Parimi, W.-T. Lu, P. Vodo, and S. Sridhar, Nature {\bf
878: 426}, 404 (2003).
879:
880: \bibitem{Sridhar2} P. V. Parimi, W.-T. Lu, P. Vodo, J. Sokoloff, J.
881: S. Derov, and S. Sridhar, Phys. Rev. Lett. {\bf 92}, 127401
882: (2004).
883:
884: \bibitem{Twersky} V. Twersky, J. Acoust. Soc. Am. {\bf 24}, 42
885: (1951); {\it ibid}, J. Math. Phys. (N.Y.) {\bf 3}, 700 (1962).
886:
887: \bibitem{JOSA} D. Felbacq, G. Tayeb, and D. Maystre, J. Opt. Soc.
888: Am. {\bf A} 11, 2526 (1994).
889:
890: \bibitem{Bikash} B. Gupta and Z. Ye, J. Appl. Phys. {\bf 94}, 2173 (2003).
891:
892: \bibitem{ChenPRL} Y.-Y. Chen and Z. Ye, Phys. Rev. Lett.
893: {\bf 87}, 184301 (2001).
894:
895: \bibitem{Yue} H. Kagemoto and D. K.-P. Yue, J. Fluid Mech. {\bf
896: 166}, 189 (1986).
897:
898: \bibitem{Tang} H.-T. Tang, private communication.
899:
900: \bibitem{Kuo2} C.-H. Kuo and Z. Ye, Phys. Rev. E ({\it in press}) ; cond-mat/0310423.
901:
902: \bibitem{Chien} H.-T. Chien, H.-T. Tang, C.-H. Kuo, C.-C. Chen,
903: and Z. Ye, Phys. Rev. B {\bf 70}, 113101 (2004).
904:
905: \bibitem{Chen} L.-S. Chen, C.-H. Kuo, and Z. Ye, Phys. Rev. E {\bf 69}, 066612 (2004).
906:
907: \bibitem{Chen5} L.-S. Chen, C.-H. Kuo, and Z. Ye, Appl. Phys. Letts. {\bf 85}, 1072 (2004).
908:
909: \bibitem{Li} Z.-Y. Li and L.-L. Lin, Phys. Rev. B {\bf 68}, 245110
910: (2003).
911:
912: \bibitem{GV} C.-H. Kuo and Z. Ye, Phys. Rev. E. 70, 046617 (2004); cond-mat/0405008.
913:
914: \bibitem{PLA} C.-H. Kuo and Z. Ye, Phys. Letts. A 331, 342 (2004).
915:
916: \bibitem{Kuo7} C.-H. Kuo and Z. Ye, J. Appl. Phys. ({\it in press}); cond-mat/0312288.
917:
918: \bibitem{Kuo3} C.-H. Kuo and Z. Ye, Phys. Rev E
919: {\bf 70}, 026608 (2004).
920:
921: \bibitem{Report3} http://physicsweb.org/article/news/8/7/9.
922:
923: \end{thebibliography}
924: \end{document}
925: