1: \documentclass[prl,aps,twocolumn,floats,showpacspsfig]{revtex4} 

2: \usepackage{epsfig}

3: %\tightenlines

4: \def\gp{\widetilde{g_\perp}}

5: \def\gpa{\widetilde{g_\parallel}}

6: \def\ggp{g_\perp}

7: \def\ggpa{g_\parallel}

8: \def\Jb{\bar{J}}

9: \def\Ib{\bar{I}}

10: \def\Kb{\bar{K}}

11: \def\be{\begin{equation}}

12: \def\ee{\end{equation}}

13: \def\bdm{\begin{displaymath}}

14: \def\edm{\end{displaymath}}

15: \def\bea{\begin{eqnarray}}

16: \def\eea{\end{eqnarray}}

17: \def\nn{\nonumber\\}

18: \def\up{\uparrow}

19: \def\da{\downarrow}

20: \def\sgn{{\rm sgn}}

21: \def\eps{\epsilon}

22: \def\r#1{(\ref{#1})}

23: \def\phib{\skew5\bar\phi}

24: \def\Phib{\bar\Phi}

25: \def\psib{\skew5\bar\psi}

26: \def\zetab{\bar\zeta}

27: \def\qb{{\bar q}}

28: \def\pb{{\bar p}}

29: \def\zb{{\bar z}}

30: \def\wb{{\bar w}}

31: \def\ab{{\bar a}}

32: \def\partialb{{\bar\partial}}

33: \def\sb{{\bar s}}

34: \def\s{\sigma}

35: \def\m{\mu}

36: \def\bra#1{\langle #1 |}

37: \def\ket#1{|#1\rangle}

38: \newcommand{\rD}{\mbox{D}}

39: \newcommand{\reff}{\mbox{eff}}

40: \newcommand{\rR}{\mbox{R}}

41: \newcommand{\rL}{\mbox{L}}

42: \newcommand{\p}{\partial}

43: \newcommand{\rF}{\mbox{F}}

44: \newcommand{\rf}{\mbox{f}}

45: \newcommand{\rc}{\mbox{c}}

46: \newcommand{\rs}{\mbox{s}}

47: \newcommand{\down}{\downarrow}

48: \newcommand{\la}{\langle}

49: \newcommand{\ra}{\rangle}

50: \newcommand{\rd}{\mbox{d}}

51: \newcommand{\ri}{\mbox{i}}

52: \newcommand{\re}{\mbox{e}}

53: \newcommand{\sumnn}{\sum_{\langle jk \rangle}}

54: \newcommand{\rk}{\mbox{k}}

55: \begin{document}

56: \draft

57: \title{Theory of Optical Transmission Through a Near-Field Probe With a 

58: Dissipative Matter in its Core}

59: \author{V. S. Lebedev, T. I. Kuznetsova  and A. M. Tsvelik $^*$}

60: \affiliation{P.N.~Lebedev Physical Institute, Leninsky prospect 53,  

61: Moscow 119991, Russia\\

62: $^*$  Department of  Physics, Brookhaven 

63: National Laboratory, Upton, NY 11973-5000, USA}

64: \date{\today}

65: 

66: 

67: \begin{abstract}

68: We develop a theory of light transmission through an  aperture-type 

69: near-field optical probe with a dissipative matter in its semiconducting core 

70: described by  a complex frequency-dependent dielectric function. 

71: We evaluate the near-field transmission coefficient of a  metallized 

72: silicon probe with a large taper angle of  

73: in the visible and near-infrared wavelength range. It is shown that in this 

74: spectral range  the use of a short silicon probe  instead of a glass one 

75: allows to achieve a  strong (up to 10$^2-10^{3}$) enhancement in the 

76: transmission efficiency. 

77: 

78: 

79: \end{abstract}

80: \pacs{ PACS No: 78.20.Bh, 78.67.-n, 42.25.Bs}

81: \maketitle

82: 

83: The development of scanning near-field optical microscopy (SNOM) has 

84: already led to enormous progress in studies of different nanoscale 

85: phenomena. Among other things, this technique is widely used for image 

86: formation of various nanoobjects, single molecule fluorescence detection,  

87: and for laser-induced ablation of a sample close to the tip apex 

88: (see \cite{Hecht2000, Girard2000}). 

89: In the recent years SNOM methods have been  employed in studies of 

90: quantum dots \cite{Kawazae2002} and single-walled carbon nanotubes 

91: \cite{Hartschuh2003}. The near-field optical microscopy has opened a 

92: possibility for direct observation of fine features in the self-focusing 

93: effect \cite{Song2000}, for imaging the light propagation in photonic 

94: crystal waveguides \cite{Bozhevolnyi2002} and the electromagnetic local 

95: density of states of optical corrals \cite{Chicanne2002}.  

96:  

97: 

98: To improve efficiency of the aperture probes 

99: for the SNOM 

100: technique one needs simultaneously to increase their transmittance  

101: and spatial resolution capability. In the present letter  we study 

102: theoretically one possible way to enhance the optical transmittance 

103: through a metallized near-field probe with a subwavelength aperture. 

104: We suggest  to work in the visible region using a short probe with a large taper angle with a core consisting 

105: of a semiconducting matter with a high refractive index $n$.  The increase in transmittance with $n$ 

106: occurs due to decrease of the light wavelength $\lambda _{c} = 2\pi c/n\omega$ inside 

107: the core. 

108: Consequently, at high $n$ 

109: the cutoff effect, which strongly reduces the light transmission through the  

110: probe in the overdamped regime, affects its transmission efficiency much 

111: less than for a glass. However, this gain  may be counterbalanced by the growth of  light absorption in the dissipative 

112: medium of the core described by the imaginary part of the 

113: frequency dependent dielectric function $\varepsilon(\omega)$. To deal 

114: with these two competing tendencies one needs a detailed theory.  

115: 

116: The purpose of this letter is to  develop an  

117: analytic theory for transmission of visible light through a probe with a 

118: core made of a dissipative semiconducting matter. Here we are encouraged by  the transmission  SNOM experiments with a silicon probe  at $\lambda = 1.06$ $\mu$m

119: \cite{Danzebrink98, Dziomba2001}, which  indicate that such probes are very 

120: promising in the near-infrared (IR) region. The theoretical 

121: support for this work comes from  a comparative numerical analysis of the transmission efficiencies 

122: of glass and silicon probes at $\lambda=1.3$ $\mu$m 

123: \cite{Castiaux98}. We have to note however, that the above theory  

124:  is restricted by the use of a two-dimensional model with 

125: a loss-free dielectric core and small taper angle of $15^{\circ}$.  

126: The further experimental success  

127: was achieved in \cite{Yatsui2002}, where the authors 

128: have employed a pyramidal Si probe that was entirely coated with a 

129: thin metal film to increase the transmission efficiency in the near-IR   

130: region ($\lambda =830$ nm). An extremely high throughput ($2.3$ \%) 

131: was obtained  in this experiment with a  resolution capability about 

132: $85$ nm. In the IR  region the light absorption in Si is 

133: sufficiently small, but  in the visible region the imaginary part of its 

134: dielectric function rapidly increases with a decrease of $\lambda$. So, 

135:  dissipation of the electromagnetic energy inside the silicon core 

136: and frequency dispersion of its dielectric function 

137: become important 

138: and should be taken into account to get an adequate physical pattern of 

139: light transmission through the probe. 

140: 

141:  The three-dimensional theory developed in the present 

142: letter is valid for all taper angles including large ones which are especially 

143: suitable for large transmittance \cite{Novotny95, KLT03}. The theory is 

144: based on the exact analytic description of the conical waveguide eigenmodes 

145: inside a probe with a dissipative matter in its core and perfectly conducting 

146: metallic walls. For a loss-free dielectric core, similar approach  has been 

147: developed in our recent work \cite{KLT03}. Here  we consider  

148: the case when the semiconducting core has a complex dielectric function 

149: which is a necessary feature of the visible region. This generalization  

150: requires more than a trivial analytic continuation of our previous results 

151: due to the necessity of describing the effects associated with frequency 

152: dispersion and light absorption inside the core matter. Such effects 

153: lead to a number of new  features in light transmission through 

154: semiconducting probes, which discuss below.   

155: 

156: 

157: 

158:  We consider here time-harmonic fields inside a cone whose core 

159: consists of a dissipative medium and whose walls are perfectly conducting. 

160: In spherical coordinates the Helmholtz equation for the Hertz function $U$ 

161: of the electromagnetic field inside a cone is  

162: 

163: \begin{equation}

164: \frac{\partial ^{2}U}{\partial r^{2}}+\frac{1}{r^{2}}\left[ \frac{1}{\sin

165: \theta }\frac{\partial }{\partial \theta }\left( \sin \theta \frac{\partial U%

166: }{\partial \theta }\right) +\frac{1}{\sin ^{2}\theta }\frac{\partial ^{2}U}{%

167: \partial \varphi ^{2}}\right] +k^{2}U=0.  \label{w-eq}

168: \end{equation}

169: 

170: \noindent Here $r$ is the distance from the cone vertex, $\theta $ and 

171: $\varphi $ are the polar and azimuthal angles. 

172: For a dissipative  matter, the wave number $k$ is complex: $

173: k=\omega \sqrt{\varepsilon \mu }/c$, $\sqrt{\varepsilon \mu }=n+i\kappa$,  

174: where  $c$ is the speed  of light, $n$ and $\kappa $ are the refractive 

175: index and the attenuation coefficient. Assuming the 

176: permeability $\mu =1$, the real 

177: $\varepsilon'=\Re e\left\{ \varepsilon \right\} $ and imaginary 

178: $\varepsilon''=\Im m\left\{ \varepsilon \right\} $ parts of the 

179: frequency-dependent dielectric function 

180: $\varepsilon \left( \omega \right) =\varepsilon'\left( \omega

181: \right) +i\varepsilon''\left( \omega \right)$ 

182: are given by $\varepsilon'=n^{2}-\kappa ^{2}$ and 

183: $\varepsilon''=2n\kappa $ (see Ref.\cite{LL}).  

184: 

185: 

186: The relevant solution of Eq. (\ref{w-eq}) corresponding  to the standing wave 

187: with vanishing amplitude at the cone vertex ($r=0$), has the form

188: 

189: \begin{equation}

190: U=\mathcal{R}\left( r\right) P_{\nu }^{m}\left( \cos

191: \theta \right) e^{im\varphi }\,, \quad

192: \mathcal{R}=Crj_{\nu }\left(kr\right) \,,

193: \label{potent}

194: \end{equation}

195: 

196: \noindent at which the radial dependence $\mathcal{R}\left( r\right) $ of 

197: the Hertz function (\ref{potent}) is expressed through the spherical Bessel 

198: function of the first kind $j_{\nu }\left( z\right)$ of a complex argument 

199: with the index $\nu $ not equal to an integer. Here $C$ is a constant. 

200: The dependence on the polar angle $\theta $ is determined by the associated 

201: Legendre function of the first kind 

202: $P_{\nu }^{m}\left( \cos \theta \right) $ with power $\nu $ and order $m$ 

203: ($m$ is an integer). 

204: 

205: For the transverse magnetic (TM) field modes 

206: the boundary condition at an interface between a core medium and a 

207: perfectly conducting metallic coating of a conical waveguide can be written 

208: as $P_{\nu }^{m}\left( \cos \theta _{0}\right) =0$, 

209: where $\theta _{0}$ is the cone half-angle. Each choice of numbers $m$ 

210: and $n$ in this equation ($n$ denotes the number of its root) 

211: determines a possible TM$_{mn}$ mode.  

212: The eigenvalues $\nu _{mn}$ strongly depend upon the value of 

213: $\theta _{0}$ such that $\nu_{mn}$ decreases with an increase of $\theta_0$. 

214: In the most interesting case of the lowest-order TM$_{01}$ mode, the 

215: projections of electric $\mathbf{E}$ and magnetic $\mathbf{H}$ fields onto 

216: the corresponding axes of spherical coordinates $(r,\,\theta ,\,\varphi )$, 

217: take the form  

218: 

219: \begin{equation}

220: E_{r}=\frac{\nu \left( \nu +1\right) }{r^{2}}\,\mathcal{R}(r)P_{\nu }\left(

221: \cos \theta \right), ~~

222: E_{\theta }=\frac{\partial \mathcal{R}(r)}{r\partial r}\frac{%

223: \partial P_{\nu }\left( \cos \theta \right) }{\partial \theta },

224: \label{E_theta}

225: \end{equation}

226: 

227: \begin{equation}

228: H_{\varphi }=i\frac{\omega \left( \varepsilon ^{\prime }+i\varepsilon

229: ^{\prime \prime }\right) }{c}\frac{1}{r}\mathcal{R}(r)\frac{\partial P_{\nu

230: }\left( \cos \theta \right) }{\partial \theta }\,.\label{H_phi}

231: \end{equation}

232: 

233: \noindent while the $H_{r}$, $H_{\theta}$, and $E_{\varphi}$ components 

234: are equal to zero.  

235: 

236: 

237: In a dissipative media with frequency-dependent dielectric function 

238: $\varepsilon =\varepsilon'+\varepsilon'' $ 

239: and permeability $\mu =\mu '+i\mu ''$, the general expressions (see 

240: \cite{LL}) for the time-averaged densities of the electric,  

241: $w_{el}=w_{r}+w_{\theta }$, and magnetic, $w_{m}=w_{\varphi }$, fields  

242: are given by

243: 

244: \begin{equation}

245: w_{el}=\frac{1}{16\pi }\frac{d\left( \omega \varepsilon '\right) }{d\omega }

246: \left( \left| E_{r}\right| ^{2}+\left| E_{\theta}\right| ^{2} \right)\,,  

247: \label{w_el}

248: \end{equation}

249: 

250: 

251: \begin{equation}

252: w_{m }=\frac{1}{16\pi }\frac{d\left( \omega \mu '\right) }

253: {d\omega }\left| H_{\varphi }\right| ^{2},  \label{w_phi}

254: \end{equation}

255:   

256: \noindent To determine the near-field transmission coefficient of a 

257: truncated conical waveguide we introduce the quantities 

258: 

259: \begin{equation}

260: W_{\beta}(r)=2\pi r^{2}\int\limits_{0}^{\theta _{0}}

261: w_{\beta}\left( r,\theta \right)

262: \sin \theta d\theta \,,   

263: \label{W_beta}

264: \end{equation}

265: 

266: \noindent which represent the integrals of 

267: $w_{r}$, $w_{\theta }$, or $w_{\varphi }$ 

268: taken over a  part of spherical surface lying inside the  cone 

269: ($0\leq \theta \leq \theta _{0},\;0\leq \varphi \leq 2\pi $)\ at a given 

270: distance $r$ from the cone vertex. With the help of Eqs. 

271: (\ref{potent}), and (\ref{E_theta})--(\ref{W_beta}), these integrals can be 

272: evaluated explicitly. 

273: The resulting expressions for $W_r$,  $W_{\theta}$, and $W_{\varphi}$, 

274: take the form

275: 

276: \begin{eqnarray}

277: && W_{r}=\frac{\left| C\right| ^{2}}{8}\frac{d\left( \omega

278: \varepsilon ^{\prime }\right) }{d\omega } \nu \left( \nu +1\right)

279: \mathcal{I}_{\nu }\left| j_{\nu }\left[

280: \left( n+i\kappa \right) \frac{\omega r}{c}\right] \right| ^{2},

281: \label{W_r}

282: \end{eqnarray}

283: \begin{eqnarray}

284:  &&W_{\theta }=\frac{\left| C\right| ^{2}}{8}\frac{d\left(

285: \omega \varepsilon ^{\prime }\right) }{d\omega }

286: \mathcal{I}_{\nu } 

287: \left| 

288: \left( \nu +1\right) j_{\nu }\left[ \left( n+i\kappa

289: \right) \frac{\omega r}{c}\right]\right . \nonumber\\

290: && \left .  -\left[ \left( n+i\kappa \right) \frac{%

291: \omega r}{c}\right] j_{\nu +1}\left[ \left( n+i\kappa \right) \frac{%

292: \omega r}{c}\right]\right| ^{2},

293: \label{W_theta}

294: \end{eqnarray}

295: \begin{eqnarray}

296: && W_{\varphi } =\frac{\left| C\right| ^{2}\left| \varepsilon \right| }

297: {8}\left( \frac{\omega r}{c}\right) ^{2}

298: \mathcal{I}_{\nu }

299: \left| j_{\nu }\left[ \left( n+i\kappa \right) \frac{\omega r}{c}\right]

300: \right| ^{2}.  \label{W_m}

301: \end{eqnarray}

302: 

303: 

304: \noindent Here the angular integral $\mathcal{I}_{\nu }$ 

305: is given by 

306: 

307: \begin{equation}

308: \mathcal{I}_{\nu }=\int\limits_{0}^{\theta _{0}}\left[ 

309: \frac{\partial P_{\nu }\left( \cos \theta \right) }{\partial \theta }\right]

310: ^{2}\sin \theta d\theta \,.  

311: \label{ang_int2}

312: \end{equation}

313: 

314: 

315: The integral energy density $W_{tot}=W_{r}+W_{\theta}+W_{\varphi}$  

316: can now be evaluated as the sum of Eqs. (\ref{W_r})--(\ref{W_m}). At small 

317: distances from 

318: the cone vertex ($r \ll \lambda_c$) it  

319: exhibits a rapid power fall, $W_{tot} \propto (|k|r)^{2\nu}$, with a 

320: decrease of $r$. At distances $r$ from the cone vertex much 

321: greater than the wavelength in the core medium ($r\gg \lambda_{c}$), 

322: the asymptotic expression for $W_{tot}$ takes the form 

323:  

324: \begin{equation}

325: \begin{array}{l}

326: W_{tot}=

327: \frac{\left| C\right| ^{2}}{16}

328: \mathcal{I}_{\nu }\left\{ \left[ 

329: \frac{d\left( \omega \varepsilon ^{\prime }\right) }{d\omega }+\left|

330: \varepsilon \right| \right] 

331: \cosh \left( \frac{r}{r_{\kappa}}\right)\right. \\ 

332: \\ 

333: +\left. \left[ \frac{d\left( \omega \varepsilon ^{\prime }\right) }{d\omega }%

334: -\left| \varepsilon \right| \right] \cos \left( 2n\frac{\omega r}{c}-\pi \nu

335: \right) \right\} .

336: \end{array}

337: \label{W_tot_as}

338: \end{equation}

339: 

340: 

341: \noindent Here $r_{\kappa}=c/2\kappa \omega $ is the attenuation length. 

342: It is evident that the main feature in the radial dependence 

343: (\ref{W_tot_as}) is determined by a factor 

344: $\cosh\left(r/r_{\kappa}\right)$, which reflects the influence of 

345: light absorption inside a dissipative core of a near-field probe. 

346: Another important point is that  the integral energy density 

347: $W_{tot}(r)$ exhibits an oscillatory behavior at distances far from the 

348: cone vertex $r\gg c/\omega \left| n+i\kappa \right|$. These additional 

349: oscillations are the result of the frequency-dependent dielectric 

350: function; they are absent if the core is made of a loss-free medium 

351: (this is the case of a probe with a glass core, 

352: $\varepsilon =const$ and $\kappa =0$, see \cite{KLT03}). 

353: However, for a lossy matter the amplitudes of the electric 

354: ($W_{el}\propto d\left( \omega \varepsilon' \right)/d\omega $,  

355: Eqs. (\ref{W_r}), (\ref{W_theta})) and magnetic 

356: ($W_{m}\propto \left| \varepsilon \right|$, 

357: Eq. (\ref{W_m})) components of the integral energy densities are not equal and 

358: can be significantly differed from each other. Therefore, the oscillations 

359: of the electric, $W_{el}$, and magnetic, $W_{m}$, energies do not 

360: compensate each other far from the cone vertex. Note also that in the 

361: presence of light absorption there is a significant 

362: difference in magnitudes of the energy fluxes associated with the 

363: incident and the reflected waves. Their ratio $\left|S_r/S_{in}\right|$ 

364: is equal to $\exp\left(-2r/r_{\kappa}\right)$. 

365: So, this difference becomes particularly important in the range of 

366: $r\gg r_{\kappa}$. 

367: 

368: 

369: Further we evaluate the optical transmittance of a conical waveguide 

370: with a dissipative matter in its core. For a probe tapered to a 

371: subwavelength diameter it is necessary to distinguish a near-field 

372: transmission coefficient of a waveguide itself and the resulting transmission 

373: coefficient to the far-field zone (see \cite{Hecht2000, KLT03}).     

374: The near-field transmission coefficient, $T$, can be expressed in terms of 

375: the time-averaged energy densities associated with the output and the input 

376: fields of the waveguide. For  spherical waves inside a cone, 

377: this coefficient 

378: can be defined as the ratio $T=W_{tot}^{out}/W_{tot}^{in} $

379: of the time-averaged energy density $W_{tot}^{out}\equiv 

380: W_{tot}\left( z_{0}\right) $ at the exit plane $z=z_{0}$ of a

381: probe integrated over the aperture cross section $2\pi \rho d\rho $ 

382: with radius $a$ ($a=z_{0}\tan \theta _{0}$) to the corresponding 

383: integral energy density $W_{tot}^{in}$ at the waveguide entrance with 

384: radial coordinate $r_{in}$. In a dissipative medium, the latter is given 

385: by $W_{tot}^{in}=\alpha W_{tot}\left(r_{in} \right)$ (see (\ref{W_beta})). 

386: The factor 

387: $\alpha =\left[1+\exp \left( -2r_{in}/r_{\kappa}\right)\right]^{-1}$  

388: shows a fraction of the integral energy density at $r=r_{in}$,  

389: associated with the incident wave alone. So, the contribution 

390: of the reflected wave turns out to be completely removed. 

391: 

392: For the subwavelength aperture $2a\ll \lambda _c$,  

393: the basic expressions (\ref{W_r})--(\ref{W_m}) for $W_{r}$, $W_{\theta}$, 

394: and $W_{\varphi}$ 

395: at the exit of a conical waveguide, can be 

396: expanded in power series of $\left|k\right|r_{out}$, where $r_{out}=a/\sin\theta_0$ is the corresponding radial coordinate. For the value of 

397: $W_{tot}\left(r_{in}\right)$ at the waveguide entrance ($r_{in}\gg \lambda_c$) 

398: we use the expression 

399: $\overline{W}_{tot}\left(r\right) \propto \cosh \left( r/r_{\kappa}\right)$,  

400: averaged over the fast oscillations of Eq. (\ref{W_tot_as}). Then, the 

401: resulting expression for the near-field transmission coefficient 

402: takes the form

403:   

404: 

405: \begin{equation}

406: T\propto 

407: \left( \frac{\omega \left|n+i\kappa \right| a}

408: {c \sin \theta_0}\right)^{2\nu\left(\theta_0\right)}

409:  \cosh^{-1} \left[\frac{ r_{in}}{r_{\kappa}\left(\omega\right)}\right]\, .

410: \label{T_anal}

411: \end{equation}

412: 

413: 

414: \noindent The eigenvalues $\nu\equiv \nu_{01}$ of the TM$_{01}$ 

415: mode in Eq. (\ref{T_anal}) exhibit rapid fall with an increase of the 

416: taper angle ($\nu_{01}=4.083$,  $2.548$, $1.777$, and $1$ at 

417: $\theta_0=\pi/6$,  $\pi/4$, $\pi/3$, and $\pi/2$, respectively). 

418: Thus, Eq. (\ref{T_anal}) describes well all major features of 

419: light transmission through the subwavelength aperture in a conical 

420: waveguide with a dissipative matter in its core. It is seen 

421: that the values of $T$ are strongly dependent on the ratio $a/\lambda$, 

422: the taper angle $2\theta_0$, and the refractive index $n$. 

423: Moreover, according to (\ref{T_anal}) the transmission coefficient 

424: $T$ is proportional to $\cosh^{-1} \xi $, where $\xi=l/r_{\kappa}$ 

425: is the ratio of the length of the probe edge $l$ to the  

426: attenuation length $r_{\kappa}=c/2\kappa \omega$ 

427: (at small $a$ we have $r_{in} \approx l$). It is clear that  the high 

428: transmission efficiency of a semiconducting probe can be achieved in the 

429: wavelength region far from the peak in its absorption band ($\kappa\ll n$). 

430: If additionally $r_{\kappa} \gg l$, than one can put 

431: $r_{\kappa}\rightarrow \infty $. Then, Eq. (\ref{T_anal}) is reduced 

432: to especially simple form $T \propto 

433: \left(\omega n a/c \sin\theta_0\right)^{2\nu\left(\theta_0\right)}$. 

434: This is the case of a loss-free dielectric core. In the opposite case 

435: of large losses ($r_{\kappa} \ll l$), the transmission coefficient   

436: behaves like $T\propto \left(\omega n a/c \sin\theta_0\right)^{2\nu\left

437: (\theta_0\right)}\exp \left(-\xi\right)$. This reflects the 

438: strong influence of light absorption on the value of $T$.  

439: 

440: 

441: Now we apply our theory for studies of the 

442: transmittance of the visible and near-IR radiation through the 

443: aperture-type metallized silicon probe. In the wavelength region from 830 nm ($\hbar\omega = 1.5$ ev)

444: down to 400 nm ($\hbar\omega = 3.1$ ev), the refractive index $n$ of Si increases monotonically 

445: from $3.67$ to $5.57$ and the attenuation coefficient $\kappa $ grows 

446: from $0.005$ to $0.387$.  

447: This results in quite different influence of light absorption in Si 

448: in the near-IR and the short-wavelength part of the visible 

449: spectrum. 

450: 

451: In Fig. 1 we present the wavelength dependences of the near-field transmission 

452: coefficient of the metallized silicon probe for the most interesting case of 

453: large taper angle $2\theta_0=90^{\circ}$ and for various values of the 

454: aperture diameter $2a$. To demonstrate a dependence  of light absorption 

455: inside the Si core on the length of the probe edge, 

456:  we calculated the transmission coefficient $T$  for various values of $l$. As expected, $T$ is strongly dependent on the aperture diameter 

457: in full agreement with simple formula (\ref{T_anal}) derived in this work. 

458: However, the wavelength dependence, obtained in the present work 

459: for the Si probe, differs dramatically from the case of a loss-free dielectric 

460: core (a glass fiber), for which $T\propto \left(a/\lambda\right)^{2\nu}$). 

461: As is evident from Fig. 1, the transmittance of the silicon probe strongly 

462: varies over the spectrum. If the length of the probe edge is not too large 

463: ($l\lesssim 10$ $\mu$m), the transmission coefficient 

464: increases first as the wavelength decreases from the IR region, 

465: reaches its maximum at a definite wavelength $\lambda_{\max}$, and then 

466: strongly falls at $\lambda \ll \lambda_{\max}$ in the short-wavelength part 

467: of the visible spectrum. The position of the maximum $\lambda_{\max}$ and the 

468: maximal value of $T_{\max}$ depends on the specific geometrical parameters of 

469: the probe. 

470: 

471: \begin{figure}[ht]

472: \begin{center}

473: \epsfxsize=0.4\textwidth

474: \epsfbox{FIG1a.eps}

475: \epsfxsize=0.4\textwidth

476: \epsfbox{FIG1b.eps}

477: \epsfxsize=0.4\textwidth

478: \epsfbox{FIG1c.eps}

479: \end{center}

480: %\label{zone}}

481: \caption{The near-field transmission coefficient, 

482: $T$, of the conical waveguide  ($2\theta _{0}=90^{\circ}$) 

483: with the Si core {\it vs}  $\lambda =2\pi c/\omega$. Curves 1, 2, 3, and 4 

484: correspond to the aperture diameter $2a=100$ nm, $70 $ nm, $50$ nm, 

485: and $25$ nm, respectively. The  

486: length of the probe edge is $l=2$ $\mu$m (a), 4 $\mu$m (b), 8 $\mu$m (c).} 

487: \end{figure}

488: 

489: 

490: It is important to stress that this maximum in the transmission efficiency 

491: of a silicon probe lies at $\lambda \sim 550-800$ nm (see Fig. 1). This occurs 

492: despite the fact that the attenuation 

493: length $r_{\kappa}$, associated with the imaginary part of the dielectric 

494: function of Si, considerably decreases in the visible region compared to the 

495: near-infrared one. For example, at $\lambda =633$ nm,  

496: $532$ nm, and $488$ nm, the respective values of the attenuation length 

497: $r_{\kappa}$ turn out to be equal to $2.66$ $\mu$m, 0.84 $\mu$m 

498: and 0.45 $\mu$m, in contrast with  

499: 13.53 $\mu$m at $\lambda=830$ nm. 

500: 

501: 

502: In summary, it follows from our calculations that at large taper 

503: angles high values of the near-field transmission coefficient 

504: can be achieved for a  passage of visible light through Si core 

505: of an optical probe. To illustrate the enhancement in the trasmittance of Si 

506: probes in comparison with conventional fiber ones we compare the present 

507: estimates for Si with those obtained for a core with small 

508: $n$ (glass or SiO$_2$). 

509: Although the taper angles of fiber probes do not usually exceed 

510: 40$^{\circ}$ which  additionally restricts their efficiency, we use 

511: the value $2\theta_0=90^{\circ}$ and $n = 1.55$ to 

512: make comparison with our recent results \cite{KLT03}.       

513: For the probe with the length $l=2$ $\mu$m and the aperture diameter 

514: $2a =50$ nm we get 

515: $T_{Si}/T_{glass} = 2.2$, 14, 45, and 71 for $\lambda = 488$, 532, 633, 

516: and 830 nm, respectively. For the same parameters, but $2\theta_0=60^{\circ}$ 

517: we have $T_{Si}/T_{glass} =43$, 240, 800, and 960. According to our theory, 

518: the enhancement  occurs as a result of competition between two factors: the 

519: rise of $n$ and the decrease of the attenuation length $r_{\kappa}$ (\ref{T_anal}). As follows from our results, the former factor is more 

520: important in the most part of the visible spectrum, provided the near-field 

521: probe length is sufficiently short (no more than several $\mu$m), such that 

522: effects associated with the light absorption are not too strong. 

523: We also would like to point out that in case of an entirely coated 

524: probe the surface plasmon-polariton propagation along the metallic 

525: cladding is likely to further increase the resulting transmittance in 

526: accordance with the  mechanism discussed in Ref. \cite{Novotny95}.  

527: 

528: 

529: This work was supported in part by the Programme ''Optical Spectroscopy 

530: and Frequency Standards'' of the Division of Physical Sciences of the Russian 

531: Academy of Sciences.   

532: AMT acknowledges the support from US DOE under contract number 

533: DE-AC02 -98 CH 10886. VSL acknowledges a support 

534: from Institute for Strongly Correlated and Complex Systems at BNL. 

535: 

536: 

537: \begin{thebibliography}{99}

538: 

539: 

540: \bibitem{Hecht2000} B.~Hecht, B.~Sick, U.~P. Wild, V.~Deckert, R.~Zenobi, 

541: O.~J.~F.~Martin, and D.~W.~Pohl, J. Chem. Phys., \textbf{112}, 7761 (2000).

542: 

543: \bibitem{Girard2000} C.~Girard, C.~Joachim, and S.~Gauthier, 

544: Rep. Prog. Phys., \textbf{63}, 893, (2000).

545: 

546: \bibitem{Kawazae2002} T.~Kawazoe, K.~Kobayashi, J.~Lim, Y.~Narita, 

547:  and M.~Ohtsu, Phys. Rev. Lett., \textbf{88}, 067404 (2002).

548: 

549: \bibitem{Hartschuh2003} A.~Hartschuh, E.J.~S\'{a}nchez, 

550: X.~S.~Xie, and L.~Novotny, Phys. Rev. Lett., \textbf{90}, 095503 (2003).  

551: 

552: \bibitem{Song2000} K.-B.~Song, J.~Lee, J.-H Kim, K.~Cho, Kim S. K. 

553: Phys. Rev. Lett., \textbf{85}, 3842 (2000). 

554: 

555: \bibitem{Bozhevolnyi2002} S.I.~Bozhevolnyi, V.S.~Volkov, 

556: T.~S\o ndergaard, 

557: A.~Boltasseva, P.~I.~Borel, and M.~Kristensen, 

558: Phys. Rev. B, \textbf{66}, 235204 (2002). 

559: 

560: \bibitem{Chicanne2002} C.~Chicanne, T.~David, R.~Quidant, J.C.~Weeber, 

561: Y.~Lacroute, E.~Bourillot, A.~Dereux, G. Colasdes Francs, C. Girard,  

562: Phys. Rev. Lett., \textbf{88}, 097402 (2002). 

563: 

564: \bibitem{Danzebrink98}  H.U.~Danzebrink, A.~Castiaux, C.~Girard, 

565: X.~Bouju, and G.~Wilkening, Ultramicroscopy, \textbf{71}, 371 (1998).

566: 

567: \bibitem{Dziomba2001} T.~Dziomba,  H.U.~Danzebrink, C.~Lehrer, L.~Frey, 

568: T.~Sulzbach, and O.~Ohlsson, J. Microsc., \textbf{202}, 22 (2001).  

569: 

570: 

571: \bibitem{Castiaux98} A.~Castiaux, H.U.~Danzebrink,  and  X.~Bouju, 

572: J. Appl. Phys., \textbf{84}, 52 (1998). 

573: 

574: \bibitem{Yatsui2002}  T.~Yatsui, K.~Isumi, M.~Kourogi, and M.~Ohtsu, 

575: Appl. Phys. Lett., \textbf{80}, 2257 (2002).

576: 

577: \bibitem{Novotny95}  L.~Novotny, D.W.~Pohl, and B.~Hecht, 

578: Opt. Lett., \textbf{20}, 970 (1995); Ultramicroscopy, \textbf{61}, 1 (1995).

579: 

580: 

581: \bibitem{KLT03}  T.~I.~Kuznetsova, V.~S.~Lebedev, and A. M. Tsvelik,

582: J. Opt. A: Pure and Appl. Opt. (\textit{submitted}).

583: 

584: \bibitem{LL}  

585: L.D. Landau and E.M. Lifshitz, \textit{Electrodynamics of Continuous Media} 

586: (Addison-Wesley, New York, 1960). 

587: 

588: %%%\bibitem{Aspnes83}  D.E.~Aspnes and A.A.~Studna, 

589: %%%''Dielectric functions and optical parameters of Si, Ge, GaAs, 

590: %%%GaSb, InP, InAs, and InSb from 1.5 to 6.0 eV, ''

591: %%%Phys. Rev. B, \textbf{27}, 985 (1983).

592: 

593: 

594: \bigskip

595: 

596: \end{thebibliography}

597: 

598: 

599: \end{document}

600: 

601: