1: % ****** Start of file apssamp.tex ******
2: %
3: % This file is part of the APS files in the REVTeX 4 distribution.
4: % Version 4.0 of REVTeX, August 2001
5: %
6: % Copyright (c) 2001 The American Physical Society.
7: %
8: % See the REVTeX 4 README file for restrictions and more information.
9: %
10: % TeX'ing this file requires that you have AMS-LaTeX 2.0 installed
11: % as well as the rest of the prerequisites for REVTeX 4.0
12: %
13: % See the REVTeX 4 README file
14: % It also requires running BibTeX. The commands are as follows:
15: %
16: % 1) latex apssamp.tex
17: % 2) bibtex apssamp
18: % 3) latex apssamp.tex
19: % 4) latex apssamp.tex
20: %
21: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb,fleq]{revtex4}
22: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
23:
24: % Some other (several out of many) possibilities
25: %\documentclass[preprint,aps]{revtex4}
26: %\documentclass[preprint,aps,draft]{revtex4}
27: %\documentclass[prb]{revtex4}% Physical Review B
28:
29: \usepackage{graphicx}% Include figure files
30: \usepackage{dcolumn}% Align table columns on decimal point
31: \usepackage{bm}% bold math
32:
33: %\nofiles
34:
35: \begin{document}
36:
37: \preprint{Published in ArXiv/Physics/Optics}
38:
39: %\title{A superlens grating theory for a stigmatic (one to one) far-field image projection of the near-field }% Force line breaks with \\
40: \title{Theory of optical imaging beyond the diffraction limit
41: with a far-field superlens}% Force line breaks with \\
42:
43:
44: \author{St\'ephane Durant}
45: \email{stephane.durant@gmail.com}
46: \author{Zhaowei Liu}%
47: \author{Nicholas Fang}
48: \altaffiliation[Present address ]{Department of Mechanical and Industrial
49: Engineering, University of Illinois at Urbana-Champaign 158 Mechanical
50: Engineering Building, MC-244 1206 West Green Street
51: Urbana, IL 61801}%Lines break automatically or can be forced with \\
52: \author{Xiang Zhang}
53: \email{xiang@berkeley.edu}
54:
55: \affiliation{%
56: 5130 Etcheverry Hall, NSF Nanoscale Science and Engineering Center
57: (NSEC)\\ University of California, Berkeley CA 94720-1740, USA}%
58: %Authors' institution and/or address\\
59:
60: %\author{Charlie Author}
61: % \homepage{http://www.Second.institution.edu/~Charlie.Author}
62: %\affiliation{
63: %Second institution and/or address\\
64: %This line break forced% with \\
65: %}%
66:
67: \date{May 23rd 2005, last revision \today }% It is always \today, today,
68: % but any date may be explicitly specified
69:
70: \begin{abstract}
71: Recent theoretical and experimental studies have shown that
72: imaging with resolution well beyond the diffraction limit can be
73: obtained with so-called superlenses. Images formed by such
74: superlenses are, however, in the near field only, or a fraction of
75: wavelength away from the lens. In this paper, we propose a
76: far-field superlens (FSL) device which is composed of a planar
77: superlens with periodical corrugation. We show in theory that when
78: an object is placed in close proximity of such a FSL, a unique
79: image can be formed in far-field. As an example, we demonstrate
80: numerically that images of 40 nm lines with a 30 nm gap can be
81: obtained from far-field data with properly designed FSL working at
82: 376nm wavelength.
83: \end{abstract}
84:
85: \pacs{78.66.-w, 42.30.Lr, 78.66.Bz, 73.20.Mf}% PACS, the Physics and Astronomy
86: % Classification Scheme.
87: %\keywords{Suggested keywords}%Use showkeys class option if keyword
88: %display desired
89:
90: \maketitle
91:
92: \section{Introduction}
93: Conventional optical lens imaging suffers from the diffraction
94: limit which originates from the loss of evanescent waves that
95: cannot travel to the far field. In the near-field, an object under
96: illumination scatters both propagating and evanescent waves.
97: Propagating waves carry the low spatial resolution information on
98: the field modulation up to $\lambda_{0}/n$, where $\lambda_{0}$ is
99: the illumination wavelength in vacuum and $n$ is the refractive
100: index of the surrounding medium. On the other hand, information on
101: modulation of the field smaller than $\lambda_{0}/n$ are carried
102: by evanescent waves.
103:
104: One promising approach of imaging beyond the diffraction limit
105: emerged with the recent proposal of superlenses\cite{Pendry00}.
106: Basically a superlens is made of a slab of material that can
107: support surface waves along the slab from electromagnetic
108: excitation. In contrast with conventional material for which
109: evanescent waves decays, using a superlens, evanescent waves can
110: be transmitted with enhanced amplitude resulting from the surface
111: wave excitation. Therefore a superlens has the ability to
112: effectively recover the evanescent components radiated by the
113: object in the image plane\cite{Smith05}\cite{Fang05}. A good image
114: can be obtained if the enhancement of evanescent waves by the
115: superlens occurs within a broadband of spatial frequency. A
116: superlens can be constructed from metamaterial with effective
117: negative index media\cite{Vesalgo68}\cite{Shelby01} consisting of
118: metallic resonators, and
119: dielectric\cite{Luo03}\cite{cubukcu03}\cite{Luo02}or
120: metallic\cite{Luo03ox} photonic crystals consisting of
121: periodically varying dielectric material. Also a superlens can be
122: built from a slab of natural material that can support surface
123: wave polaritons\cite{Fang05}\cite{Stockman05}\cite{Shalaev05} with
124: either negative permittivity or negative permeability.
125: Experimental studies have demonstrated superlens imaging both in
126: microwave regime\cite{Parimi03} using a two dimensional photonic
127: crystal and in optical regime using a silver
128: film\cite{Fang05}\cite{Lee05}\cite{Melville05}. Although the
129: resolution of the superlens is still limited by the internal
130: absorption losses of the materials due to inherent in strongly
131: dispersive material\cite{Garcia02}, imaging well beyond the
132: diffraction limit has been shown.
133:
134: However, there is one drawback to the planar superlens. The
135: superlens images are still in near-field zone as demonstrated by
136: Podolskiy et al.\cite{Podolskiy05}, since the enhanced evanescent
137: waves remain evanescent after transmission and vanish very quickly
138: outside the superlens. Therefore, a challenge remains as how to
139: use the superlens effect to form an image in far field with a
140: resolution beyond the diffraction limit.
141:
142: Another approach to recover the evanescent waves is to introduce
143: an antenna into the near field that interacts with the evanescent
144: waves and then radiates into the far field. In fact, this is the
145: fundamental principle of the near-field scanning optical
146: microscopy (NSOM) where optical nanoantennas such as plasmonic
147: nanoparticles or metallic tips are used. Considerable studies have
148: been devoted to interpret the far-field
149: signals\cite{Greffet97}\cite{carminati00}\cite{Greffet95}\cite{Porto00}\cite{Carney04}
150: depending of the NSOM configuration, and images with resolution
151: down to 10nm have been demonstrated possible. Nevertheless, NSOM
152: do not project a physical image as a conventional lens does, and
153: the slow scan speed prevents a dynamic imaging\cite{Hecht00},
154: often of practical importance.
155:
156: In this letter, we show theoretically that a new device termed as
157: {\it far-field superlens} (FSL) can overcome the near-field
158: limitation of a superlens, that is, in other words, able to
159: project in far-field, an image of the evanescent component of the
160: near-field. Moreover, we demonstrate theoretically that the image
161: is unique. The image pattern is not a real space image, but rather
162: provides the field angular spectrum\cite{Goodman96}, i.e.
163: information on the object in spatial spectral domain. The
164: far-field signal can be easily processed numerically in order to
165: obtain a real space image of the local field distribution above
166: the object with a resolution beyond the diffraction limit. As an
167: example, a realistic design of an optical FSL made of
168: metal/dielectric is proposed from exact numerical calculation.
169:
170: \section{Imaging theory with a Far-field superlens made of arbitrary material}
171: Adding a periodic grating on a superlens positioned in the
172: near-field above an object may help to realize a lens-like imaging
173: with a resolution below the diffraction limit. However, the
174: imaging capability of a grating is not straightforward. Let us
175: first introduce some general transmission properties of a grating
176: without considering the superlens effect. We assume an object
177: radiating optical waves at a wavelength $\lambda_{0}$ below a
178: grating immersed into the same medium with a refractive index .
179: For the sake of simplicity and without losing generality, we
180: consider a $2$ dimensional problem where the material properties
181: of both object and grating are function of $(x,z)$ and invariant
182: along the $y$ axis. The grating is periodic along the $x$ axis
183: with a periodicity $d$. Periodic gratings are known to be able to
184: couple out evanescent waves into propagating waves by a simple
185: diffraction process. This property can be understood by writing
186: the for instance the grating law:
187: \begin{equation}
188: k'=k+pG,
189: \end{equation}
190: where $k'$ and $k$ are the transmitted and the incident
191: transverse wave number respectively; $p$ is the diffraction order;
192: and $G$ is the grating wave number of the grating. The transverse
193: wave number is the projection of the wavevector of a plane wave
194: along the $x$ axis. Transverse wave number of evanescent waves are
195: such that $|k|>nk_{0}$ where $k_{0}=2\pi/ \lambda_{0}$, while
196: transverse wave number of propagating waves satisfy to
197: $|k|<nk_{0}$. Incident evanescent waves with a large $k$ can be
198: lowered by the grating wave number using for instance the order
199: $p=-1$ of diffraction. Evanescent waves can be converted by this
200: way into propagating waves that reach the far-field if $G$ is
201: large enough. But incident propagating waves would be also
202: transmitted in far-field through the order $p=0$ without wave
203: number change. So that incident propagating and evanescent waves
204: transmitted through the order $0$ and $-1$ respectively will
205: overlap in far-field making it difficult to separate them for
206: imaging purposes. Indeed, waves transmitted in far-field for
207: instance with a transverse wave number $|k'|<nk_{0}$ are the
208: results of the overlap of incident waves transmitted through
209: several orders $p$ with transverse wave numbers satisfying:
210: \begin{equation}
211: k_{p}=k'-pG.
212: \end{equation}
213:
214: Let us write the relationship between the field transmitted in
215: far-field and the incident field assuming TM polarized waves where
216: the magnetic field $H$ is oriented along the $y$ axis. The
217: $H$-field transmitted above the grating with and its angular
218: spectrum\cite{Goodman96} are noted $H_{t}(x,z)$ and
219: $\widetilde{H}_{t}(k',z)$. Only plane waves with $|k'|<nk_{0}$
220: that can reach the far-field are considered. The near field
221: radiated by the object under the grating with $z_{0}<z<z_{1}$ is
222: noted $H_{obj}(x,z)$. The near-field can be decomposed into a
223: broadband angular spectrum of both propagating ($|k|<nk_{0}$) and
224: evanescent ($|k|>nk_{0}$) plane waves. Separated by a grating,
225: those two angular spectra are linked by a discrete linear
226: summation of waves scattered into all orders of diffraction:
227: \begin{equation}{\label{eq:transmission_}}
228: \widetilde{H}_{t}(k',z_2)=\sum_{p=-\infty}^{+\infty}
229: t_{p}(k_{p})\widetilde{H}_{obj}(k_{p},z_1),
230: \end{equation}
231: In Eq. (3), $t_{p}$ is the $p$-order field transfer function of
232: the grating from $z=z_{1}$ to $z=z_{2}$, defined as the ratio
233: between the field transmitted in the order $p$ of diffraction, and
234: the field of an incident plane wave. The transfer function is a
235: convenient tool commonly used in Fourier optics\cite{Goodman96} to
236: describe the transmission properties of optical system by plane
237: waves. Transfer functions can be either measured experimentally or
238: numerically by solving Maxwell's equations.
239:
240: In general, the original near-field $\widetilde{H}_{obj}(k,z_1)$,
241: cannot be retrieved unambiguously from the far-field measurement
242: of the angular spectrum $\widetilde{H}_{t}(k,z)$ using Eq. (2-3)
243: because of an overlap of several incident plane waves with
244: different $k_{p}$ scattered into with the same transverse wave
245: number $k'$ ( the same direction). In general, there is no
246: one-to-one relationship between the near-field angular radiated by
247: the object and the far-field angular spectrum transmitted by a
248: grating.
249:
250: \begin{figure}
251: \includegraphics[width=6.5cm]{Superlens_versus_FSL_PRL_testmod_v6_acro.eps}% Here is how to import EPS art
252: \caption{\label{fig:1} Schematic picture of the transmission
253: properties of a conventional superlens versus a far-field
254: superlens. Through a conventional superlens (a), incident
255: evanescent waves are enhanced in transmission and vanish quickly
256: in the near-field zone. In contrast, a FSL (b) both enhances and
257: converts them into propagating waves by diffraction while blocking
258: incident propagating waves.}
259: \end{figure}
260:
261: We demonstrate that this problem of the overlap of waves
262: transmitted through several order of diffraction can be overcome
263: by combining the superlens effect to the diffraction properties of
264: a grating. The imaging principles can be understood following a
265: very simple picture. Let us look first at transmission properties
266: of a planar superlens as show in Fig 1a. Transmitted amplitude of
267: incident evanescent waves (in black) are substantially enhanced
268: through the slab because of the superlens effect. Comparatively
269: incident propagating waves are poorly transmitted (in red).
270: However, after transmission enhanced evanescent waves remain
271: evanescent, limiting imaging with a superlens in the near-field
272: zone. In contrast, let us consider now transmission properties of
273: planar superlens corrugated with a subwavelength periodic
274: structure termed as {\it far-field superlens} (FSL). As shown in
275: Fig. 1b, a FSL not only enhances the incident evanescent field
276: because of the excitation of surface waves in the slab based
277: superlens, but also effectively convert these waves into
278: propagating waves by scattering through a negative diffraction
279: order of the grating following Eq 1. Overall, these incident
280: evanescent waves transmitted and converted into propagating waves
281: are projected in the far-field with large amplitude. In the others
282: hand compared to the transmission of incident evanescent waves,
283: incident propagating waves are expected to be very poorly
284: transmitted in far-field because of a lack of surface wave
285: excitation, This property may be written:
286: \begin{equation}\label{eq:fsl_condition}
287: |t_{0}(k-G)|<<|t_{-1}(k)|,
288: \end{equation}
289: with $|t_{0}(k-G)|<<|t_{-1}(k)|$ within the bandwidth of
290: evanescent waves for which the superlens effect occurs. Let us
291: note that a similar relation occurs for negative transverse wave
292: numbers if the grating has a -axis symmetry grating. If in
293: addition, the superlens is designed with a large transmission
294: within selective bandwidth $k\in [G;nk_{0}+G]$, then the
295: relationship between the far-field angular spectrum above the FSL
296: and the near-field angular spectrum below the superlens given by
297: Eq (2) and (3) reduces to:
298: \begin{equation}
299: \widetilde{H}_{t}(k',z_2)=t_{-1}(k)\widetilde{H}_{obj}(k,z) \text{
300: where } k=k'+G,
301: \end{equation}
302: for the positive half-space $0<k'<nk_{0}$; and
303: \begin{equation}
304: \widetilde{H}_{t}(k',z_1)=t_{+1}(k)\widetilde{H}_{obj}(k,z) \text{
305: where } k=k'-G,
306: \end{equation}
307: It follows from this result that any propagating wave transmitted
308: in far-field by a FSL has a unique origin. For a positive $k'$ for
309: instance, the origin is the incident evanescent wave that has been
310: transmitted through the diffraction order $-1$ with $k=k'+G$. This
311: property is true for any $k'$ so that there is a unique one to one
312: relationship between the near-field angular spectrum under the FSL
313: and the transmitted angular spectrum in far-field above the FSL.
314: This results means that when an object is placed in close
315: proximity of a FSL, a unique image of the near-field distribution
316: can be projected in far-field. Moreover, using Eq. (5) and (6) and
317: the rigorous diffraction theory\cite{Goodman96}, the near-field
318: angular spectrum radiated by the object can be retrieved
319: unambiguously from measurement of the far-field transmitted
320: angular spectrum $\widetilde{H}_{t}(k',z)$.
321:
322: If both amplitude and phase of the angular spectrum can be
323: measured in far-field, then a real space image of the near-field
324: $\widetilde{H}_{obj}(k,z)$ above the object can be reconstructed
325: from $H_{obj}(x,z)$ using a simple inverse Fourier transform.
326: However, the measurement of the phase is a practical difficulty.
327: This difficulty appears also in diffraction optical
328: microscopy\cite{Lauer02} where both amplitude and phase of the
329: angular spectrum have to be measured. For this purpose, an
330: experimental set-up such as the one use by Lauer\cite{Lauer02}
331: based on interferometry may be a good approach. Alternatively, a
332: direct real space image might be obtained using the Fourier
333: transform transmission properties of lens\cite{Goodman96} and
334: other optical devices.
335:
336: In principle, the maximum spatial frequency of the electromagnetic
337: field that a FSL can image in far-field is
338: $(n+\lambda_{0}/d)k_{0}$. Consequently, the best transverse
339: resolution $\Delta l$ that could be obtained on the image of the
340: local density of electromagnetic energy is:
341: \begin{equation}
342: \Delta l=\frac{\lambda_{0}}{2(n+\lambda_{0}/d)}.
343: \end{equation}
344: By comparison, the best resolution that could be achieved with a
345: diffraction limited microscope is $\lambda_{0}/2n$ assuming a
346: numerical aperture NA=n.
347:
348: Using a FSL, we have demonstrated that the near-field angular
349: spectrum and subsequently the local near-field distribution can be
350: measured. However, the electromagnetic distribution of the field
351: above the object depends on how the object is exposed. For
352: instance, in normal incidence or with a grazing angle exposure by
353: a plane wave, the FSL would provide accordingly different images.
354: A model is needed if one wants to image an intrinsic property of
355: the object that does not depend on the exposure condition such as
356: the local polarizability or the local absorptivity.
357:
358: \section{Case of a silver far-field superlens}
359: \begin{figure}
360: %\includegraphics[width=8.5cm]{transfer_function_plus_design_FIG3_illus_acro}% Here is how to import EPS art
361: \includegraphics[width=8.5cm]{FIG2_illus_acro2}% Here is how to import EPS art
362: \caption{\label{fig:2} (a) Design of an optical FSL working at
363: $\lambda_{0}=376nm$ with $a=45nm$ ; $b=35nm$ ; $c=55nm$ ;
364: $d=150nm.$ (b) Amplitude of transmission factor through order
365: $p=0$ (red) and order $p=-1$ (black) from near-field $z=z_{1}$ to
366: far-field ($z>>\lambda_{0}$) of the optical FSL shown in (a). This
367: FSL satisfies the requirement for imaging purpose: it provides a
368: strong transmission of evanescent waves and convert them into
369: propagating waves through the order -1 while the transmission of
370: propagating waves through the order 0 is comparatively small.}
371: \end{figure}
372:
373: How to design such a FSL for which the transmission properties
374: satisfy to Eq (4) is a crucial question. One may start from the
375: design of a superlens slab that enhances strongly incident
376: evanescent waves within a large bandwidth. The enhancement can be
377: provided by the excitation of surface waves mode of the slab based
378: superlens. When a periodic corrugation is added on a superlens,
379: the surface modes supported by the slab become leaky modes. As a
380: result, as it was demonstrated by Smith et al\cite{Smith03} in
381: case of a superlens made of metamaterial with a negative
382: refractive index, corrugations at the interfaces of the superlens
383: lead to smaller values of the enhancement of evanescent waves by
384: the superlens. Despite this expected difficulty, we have
385: successfully designed an optical FSL made of silver/glass with the
386: proper transfer functions satisfying to Eq (4), a necessary
387: condition for imaging purpose. Details on the design of this FSL
388: are provided in Ref.\cite{DURANT06}. Feature sizes of this
389: nanostructure are shown in Fig 2a. This FSL has been designed to
390: work at $\lambda_{0}=376nm$ with TM polarized waves. We have
391: computed the transfer functions of this structure from $z=z_{1}$
392: to the top $z=z_{2} $ by solving numerically Maxwell's equations
393: using the Rigorous Coupled Wave Analysis (RCWA)
394: algorithm\cite{Moharam95_1}\cite{Moharam95_2} with experimental
395: permittivity data of glass $\epsilon=2.31$ and
396: silver\cite{Johnson72} $\epsilon=-3.16+0.2i$. The numerical
397: solution provided has been tested using the theorem of reciprocity
398: of electromagnetic waves and was applied for both propagating and
399: evanescent waves\cite{Carminati00_2}\cite{Carminati98}. The
400: results of order $0$ and $\pm 1$of the amplitude transfer
401: functions are plotted in Fig. 2b. With a periodicity $d=150nm$ and
402: the wavelength $\lambda_{0}=376nm$, incident transverse wave
403: numbers transmitted through the order $-1$ of the grating are
404: shifted by $-2.5k_{0}$.
405:
406:
407: Fig. 2b clearly shows that Eq (4) is satisfied with
408: $k\in[2.5;4]k_{0}$, demonstrating that using a superlens
409: periodically corrugated, a large bandwidth of evanescent waves can
410: be both enhanced and converted into propagating waves with large
411: amplitude, while incident propagating waves are poorly
412: transmitted. Consequently, this FSL could be used for imaging with
413: resolution well below the diffraction limit. Fig 2b shows
414: similarly that $|t_{0}(k+G)|<<|t_{+1}(k)|$ with
415: $k\in[-4;-2.5]k_{0}$ .
416:
417: In a superlens made of silver, surface plasmon
418: polaritons\cite{Barnes03} (SPP) play a key role on the enhancement
419: of evanescent waves \cite{Smith05}\cite{Fang05}\cite{Lee05}. At a
420: metal/dielectric interface, SPP are surface waves resulting from
421: the coupling between p-polarized electromagnetic waves and the
422: induced collective excitation of free electrons in the metal. In a
423: silver film superlens, the wavelength and the thickness are chosen
424: so that SPP can be excited within a large bandwidth of transverse
425: wave numbers\cite{Fang05}\cite{Lee05}. How the optical FSL
426: presented in this letter has been designed in close connection to
427: SPP behavior, is detailed in Ref.\cite{DURANT06}.
428:
429: Due to the position of the selective bandwidth of enhancement as
430: shown Fig. 2b, waves transmitted into order $-1$ and $+1$ can be
431: substantially overlapped with $k\in[-0.2;0.2]$. For this reason,
432: this small bandwidth has to be omitted from the measurement in
433: order to retrieve the near-field angular spectrum unambiguously.
434: Finally, it can be deduced using Eq (5) and (6) that the near
435: field angular spectrum $\widetilde{H}_{obj}(k,z)$ with
436: $k\in[-4;-2.7]\cup[2.7;4]k_{0}$ can be retrieved from the
437: measurement of the far-field angular spectrum
438: $\widetilde{H}_{t}(k',z)$ with
439: $k'\in[-1.5;-0.2]\cup[0.2;1.5]k_{0}$. Because this specific FSL
440: can resolve a transverse field modulation with a maximum spatial
441: frequency $4k_{0}$ , the transverse resolution on the image of the
442: local density of electromagnetic energy is $\lambda_{0}/8$. By
443: comparison, the resolution of diffraction limited microscope is
444: $\lambda_{0}/3$ with a numerical aperture $NA=1.5$.
445:
446:
447: \begin{figure}
448: \includegraphics[width=8.5cm]{imagereconstruction_v4_illus_acro}% Here is how to import EPS art
449: \caption{\label{fig:3} Electromagnetic energy 5nm above the object
450: and corresponding images with and without FSL assuming in both
451: case a numerical aperture NA=1.5. Image of the density of
452: electromagnetic energy with the FSL is reconstructed using Eq 5
453: and 6 and from the rigorous computation of the transmitted angular
454: spectrum in far-field $z>>\lambda_{0}$. This result computed
455: rigorously directly demonstrates the optical imaging method with
456: resolution below the diffraction limit from Far-field data, using
457: a FSL made of silver/glass without any scanning}
458: \end{figure}
459:
460: As an example, we provide the result of the image reconstruction
461: using a FSL, of an object constituted of two 40nm lines sources of
462: coherent TM waves separated by a deep subwavelength gap. A unity
463: value of the $H$ component is assumed on the two-line sources and
464: vanished everywhere else. The near field computed at 5nm above the
465: object using rigorous diffraction theory\cite{Goodman96}, is shown
466: as the black curve in Fig. 3. For comparison, the computed
467: diffraction limited image using conventional optical microscope
468: assuming a numerical aperture $NA=1.5$ is shown in blue in Fig 3.
469: The optical FSL described in Fig 2a is placed 20nm above the
470: object. The angular spectrum $\widetilde{H}_{t}(k',z)$ transmitted
471: by the FSL in far-field ($z>>\lambda_{0}$) is computed rigorously
472: using RCWA and Eq. (2-3). Values of $\widetilde{H}_{t}(k',z)$
473: provide a complete set of data simulating a measured signals in an
474: experiment. These known data are used subsequently for the image
475: reconstruction. Because the designed silver FSL processes a unique
476: one to one $k'\rightarrow k$ relation, it allows us to use this
477: "experimental" data to retrieve the angular spectrum of the
478: near-field 5nm above the object unambiguously only by using
479: Eq.(5-6) and the rigorous diffraction theory. By combining this
480: angular spectrum with its propagating component, we obtain the
481: near-field angular spectrum 5nm above the object with
482: $k\in[-4;-2.7]\cup[-1.5;1.5]\cup[2.7;4]k_{0}$. By applying a
483: simple inverse Fourier transform, we finally obtain successively a
484: reconstruction of the $H$-field distribution and the image of the
485: density of electromagnetic energy 5nm above the object. We have
486: successively obtained faithful images reconstruction from 120nm
487: down to 30nm gap. The case of 40nm gap is reported by the red
488: curve in Fig. 3 where the separation of the two lines source is
489: very clear. Let us note the formation of some
490: artefacts\cite{DURANT06} in the image may appear because of the
491: missing band in the near-field angular spectrum. The image in case
492: of 30nm gap (not shown) is the smallest gap between sources that
493: we have been able to demonstrate following the Rayleigh
494: criterion\cite{Born80}.
495:
496: \section{Conclusion}
497: We have demonstrated theoretically how to overcome the limitation
498: of a conventional superlens for which only images in the
499: near-field can be obtained\cite{Fang05}\cite{Podolskiy05}. We have
500: shown that when the object is positioned close to a new device
501: termed as the far-field superlens (FSL), a unique image of
502: evanescent waves radiated by the object can be formed in
503: far-field. In contrast to conventional near-field scanning optical
504: microscope (NSOM), the FSL does not require scanning. In this
505: sense, a FSL is similar with conventional lenses imaging with
506: which a whole and unique image of an object can be recorded in a
507: single snap-shot. From the measurement of the far-field image
508: pattern and a simple inversion of the linear and scalar Eqs. (5)
509: and (6), the near-field electromagnetic distribution above the
510: object can be obtained with a resolution beyond the diffraction
511: limit. By combining the superlens effect and the diffraction modes
512: of a grating, the unique transmission properties of a FSL lies in
513: a broadband excitation of surface wave leaky modes used to convert
514: the incident near-field angular spectrum into a transmitted
515: far-field angular spectrum, related by a one to one relationship.
516: A realistic design of an optical FSL was given made of
517: silver/glass with such a transmission properties, owing to the
518: excitation of surface plasmon polariton (SPP) leaky modes. This
519: new imaging approach has the potential to reach similar or better
520: resolution than NSOM after more development. Such a far field
521: superlens could have great impact not only in nano-scale imaging
522: but also in nanolithography and photonic devices.
523:
524: \begin{acknowledgments}
525: We are very grateful to Dr Q.-H. Wei for the critical reading of
526: the manuscript. This research was supported by NSF Nano-scale
527: Science and Engineering Center (DMI-0327077) and ONR/DARPA
528: Multidisciplinary University Research Initiative (MURI)
529: (Grant\#N00014-01-1-0803).
530: \end{acknowledgments}
531:
532:
533: \newpage %Just because of unusual number of tables stacked at end
534: \bibliography{FSL}% Produces the bibliography via BibTeX.
535:
536: \end{document}
537: