1: %\documentclass[showpacs,preprint,amsmath,amssymb]{revtex4}

2: \documentclass[twocolumn,prb,showpacs]{revtex4}

3: 

4: 

5: %\documentclass[showpacs,preprint,amsmath,amssymb]{revtex4}

6: %\documentclass[showpacs,preprintnumbers,amsmath,amssymb,twocolumn,prb]{revtex4}

7: %\newcommand{\Sc}{Schr\"odinger }

8: % \newcommand{\al}{\alpha }

9: % \newcommand{\vfi}{\varphi }

10: 

11: %\ifx\pdftexversion\undefined

12: %  \usepackage[dvips]{graphics}

13: %\else

14: %  \usepackage[pdftex]{graphics}

15: %\fi

16: 

17: %\usepackage{graphicx}

18: 

19: 

20: 

21: \usepackage{graphicx}

22: \usepackage{epsfig}

23: \usepackage{dcolumn}

24: \usepackage{bm}

25: \usepackage{amsmath}

26: 

27: \begin{document}

28: \title{Hall Voltage with the Spin Hall Effect}

29: \author{Yuriy V. Pershin}

30: \email{pershin@physics.ucsd.edu}

31: \author{Massimiliano Di Ventra}

32: \email{diventra@physics.ucsd.edu} \affiliation{Department of

33: Physics, University of California, San Diego, La Jolla, California

34: 92093-0319}

35: 

36: \begin{abstract}

37: The spin Hall effect does not generally result in a charge Hall

38: voltage. We predict that in systems with inhomogeneous electron

39: density in the direction perpendicular to main current flow, the

40: spin Hall effect is instead accompanied by a Hall voltage. Unlike

41: the ordinary Hall effect, we find that this Hall voltage is

42: quadratic in the longitudinal electric field for a wide range of

43: parameters accessible experimentally. We also predict spin

44: accumulation in the bulk and sharp peaks of spin-Hall induced charge

45: accumulation near the edges. Our results can be readily tested

46: experimentally, and would allow the electrical measurement of the

47: spin Hall effect in non-magnetic systems and without injection of

48: spin-polarized electrons.

49: \end{abstract}

50: 

51: \pacs{72.25.Dc, 71.70.Ej}

52: 

53: \maketitle

54: 

55: Currently, much attention is given to studies of the spin Hall

56: effect, which allows to polarize electron spins without magnetic

57: fields and/or magnetic materials

58: \cite{natureSHC,r0,r1,r2,r3,Rashba,Rashba1,r4,r5,r6,r7,r8,r9,r10,Kato,Wunderlich,Valenzuela}.

59: In the spin Hall effect, electrically induced electron spin

60: polarization accumulates near the edges of a channel and is zero

61: in its central region. This effect is caused by deflection of

62: carriers moving along an applied electric field by

63: extrinsic\cite{r1} and/or intrinsic\cite{r3} mechanisms. In a

64: non-magnetic homogeneous system, spin accumulation is {\em not}

65: accompanied by a charge Hall voltage because two spin Hall

66: currents cancel each other \cite{natureSHC}. The absence of Hall

67: voltage leads to difficulties in probing the spin Hall effect,

68: since measuring a charge accumulation is much easier than

69: measuring a spin accumulation. Recently, the spin Hall effect has

70: been observed both optically \cite{Kato,Wunderlich} and

71: electrically \cite{Valenzuela}. In the latter case, a charge

72: accumulation has been created through injection of spin-polarized

73: electrons into the sample \cite{Valenzuela}.

74: 

75: In the present Letter, we predict that in a system with an

76: inhomogeneous electron density profile in the direction

77: perpendicular to the direction of main current flow, the spin Hall

78: effect results in {\em both} spin and charge accumulations. The

79: pattern of charge accumulation is determined by the interplay of two

80: mechanisms. The first mechanism of charge accumulation is based on

81: the dependence of spin-up and spin-down currents on local spin-up

82: and spin-down densities. Spin currents, outgoing from regions with

83: higher densities, are not fully compensated by incoming currents,

84: therefore, a charge accumulation appears. This mechanism is

85: primarily responsible for the non-zero Hall voltage. The second

86: mechanism of charge accumulation is related to scattering of spin

87: currents on sample boundaries which act like obstacles. Like in the

88: case of Landauer resistivity dipoles \cite{Landauer}, this

89: scattering leads to formation of local charge accumulation, which is

90: also expected in traditionally-studied spin Hall systems. In

91: addition, we show that in systems with inhomogeneous electron

92: density the spin accumulation appears not only near the sample

93: boundaries, but also in the bulk. Our approach does not involve any

94: use of magnetic materials and fields, therefore, the spin Hall

95: effect can be measured electrically in completely non-magnetic

96: system and without injection of spin-polarized electrons.

97: 

98: To illustrate this effect, let us begin by considering a system

99: having a step profile of electron density, as shown in Fig.

100: \ref{fig1}. There are several possible ways to fabricate such a

101: system including density depletion by an electrode, inhomogeneous

102: doping \cite{YP}, or variation of the sample height. What is

103: important for us is that the perpendicular (in $y$ direction) spin

104: currents are different in the regions with different electron

105: density. Then, if we consider currents passing through the boundary

106: separating regions with different charge densities ($n_1$ and

107: $n_2$), it is clear that the spin current from the region with

108: higher electron density has a larger magnitude than the current in

109: the reverse direction. The difference in currents implies charge

110: transfer through the boundary and formation of a dipole layer.

111: 

112: \begin{figure}[b]

113: \includegraphics[angle=270,width=7.5cm]{fig1}

114: \caption{\label{fig1} (Color online) Spin Hall effect in a system

115: with a step profile of the electron density in $y$ direction,

116: $n_1>n_2$. Spin currents through the boundary between $n_1$ and

117: $n_2$ do not cancel each other, resulting in a Hall voltage.}

118: \end{figure}

119: 

120: Let us now provide a quantitative analysis of this effect. We employ

121: a two-component drift-diffusion model \cite{flatte,PDV}, and in

122: order to find a self-consistent solution, we supplement the

123: drift-diffusion equations with the Poisson equation. In our

124: drift-diffusion calculation scheme, the inhomogeneous charge density

125: profile $n(y)$ is defined via an assigned positive background

126: density profile $N(y)$ (such as the one in Fig.~\ref{fig1}), which,

127: as discussed above, can be obtained in different ways. Assuming

128: homogeneous charge and current densities in $x$ direction and

129: homogeneous $x$-component of the electric field in both $x$ and $y$

130: directions, we can write a set of equations including only $y$ and

131: $t$ dependencies:

132: 

133: \begin{equation}

134: e\frac{\partial n_{\uparrow (\downarrow)}}{\partial

135: t}=\textnormal{div}  j_{y,\uparrow

136: (\downarrow)}+\frac{e}{2\tau_{sf}}\left(n_{\downarrow

137: (\uparrow)}-n_{\uparrow (\downarrow)} \right), \label{contEq}

138: \end{equation}

139: \begin{equation}

140: j_{y,\uparrow (\downarrow)}=\sigma_{\uparrow (\downarrow)}

141: E_y+eD\nabla  n_{\uparrow (\downarrow)}\pm \gamma I_{x,\uparrow

142: (\downarrow)} , \label{currentEq}

143: \end{equation}

144: and

145: \begin{equation}

146: \textnormal{div}E_y=\frac{e}{\varepsilon\varepsilon_0}\left(

147: N(y)-n\right), \label{puaeq}

148: \end{equation}

149: where $-e$ is the electron charge, $n_{\uparrow (\downarrow)}$ is

150: the density of spin-up (spin-down) electrons, $j_{y,\uparrow

151: (\downarrow)}$ is the current density, $\tau_{sf}$ is the spin

152: relaxation time, $\sigma_{\uparrow (\downarrow)}=en_{\uparrow

153: (\downarrow)}\mu$ is the spin-up (spin-down) conductivity, $\mu$ is

154: the mobility, $D$ is the diffusion coefficient, $\epsilon$ is the

155: permittivity of the bulk, and $\gamma$ is the parameter describing

156: deflection of spin-up (+) and spin-down (-) electrons. The current

157: $I_{x,\uparrow (\downarrow)}$ in $x$-direction is coupled to the

158: homogeneous electric field $E_0$ in the same direction as

159: $I_{x,\uparrow (\downarrow)}=en_{\uparrow (\downarrow)}\mu E_0$. The

160: last term in Eq. (\ref{currentEq}) is responsible for the spin Hall

161: effect.

162: 

163: Equation (\ref{contEq}) is the continuity relation that takes into

164: account spin relaxation, Eq. (\ref{currentEq}) is the expression for

165: the current in $y$ direction which includes drift, diffusion and

166: spin Hall effect components, and Eq. (\ref{puaeq}) is the Poisson

167: equation. It is assumed that $D$, $\mu$, $\tau_{sf}$ and $\gamma$

168: are equal for spin-up and spin-down electrons.~\cite{prec} In our

169: model, as it follows from Eq. (\ref{currentEq}), the spin Hall

170: correction to spin-up (spin-down) current (the last term in Eq.

171: \ref{currentEq}) is simply proportional to the local spin-up

172: (spin-down) density. All information about microscopic mechanisms

173: for the spin Hall effect is therefore lumped in the parameter

174: $\gamma$.

175: 

176: Combining equations (\ref{contEq}) and (\ref{currentEq}) for

177: different spin components we can get the following equations for

178: electron density $n=n_{\uparrow}+n_{\downarrow}$ and spin density

179: imbalance $P=n_{\uparrow}-n_{\downarrow}$:

180: \begin{equation} \frac{\partial

181: n}{\partial t}=\frac{\partial}{\partial y} \left[ \mu n E_y + D

182: \frac{\partial n}{\partial y} +\gamma P \mu E_0 \right] \label{CC}

183: \end{equation}

184: and

185: \begin{equation} \frac{\partial P}{\partial t}=\frac{\partial}{\partial y}

186: \left[ \mu P E_y + D \frac{\partial P}{\partial y} +\gamma n \mu

187: E_0 \right]-\frac{P}{\tau_{sf}}. \label{Peq}

188: \end{equation}

189: 

190: {\it Analytical solution --} Before solving Eqs.

191: (\ref{puaeq})-(\ref{Peq}) numerically, let us try to find analytical

192: solutions in specific cases. This will help us in the discussion of

193: the numerical results. An analytical steady-state solution of these

194: equations can indeed be found for the case of exponential density

195: profile in a system which is infinite in the $y$ direction.

196: 

197: The structure of Eqs. (\ref{puaeq})-(\ref{Peq}) allows us to

198: select a solution in the form

199: \begin{eqnarray}

200: n=N(y)=Ae^{\alpha y} \label{s1}, \\ P=Ce^{\alpha y},

201: \\E_y=\textnormal{const},\label{s2}

202: \end{eqnarray}

203: where $A$, $C$ and $\alpha$ are constants ($A$ and $\alpha$ are

204: assigned). This solution corresponds to constant spin polarization

205: $p=P/n$. Substituting Eqs. (\ref{s1})-(\ref{s2}) into Eqs.

206: (\ref{CC}) and (\ref{Peq}) (note that the Poisson equation

207: (\ref{puaeq}) is automatically satisfied) we obtain

208: \begin{eqnarray}

209: \mu E_y A+D\alpha A+\gamma \mu E_0C=0,\label{eq123}

210: \\ \mu E_y \alpha

211: C+D\alpha^2C+\gamma \mu E_0 \alpha A-\frac{C}{\tau_{sf}}=0.

212: \end{eqnarray}

213: From these equations, eliminating $E_y$, we find

214: \begin{equation}

215: C=\frac{-1\pm\sqrt{1+\left( 2\tau_{sf}\gamma \mu E_0 \alpha

216: \right)^2}}{2\tau_{sf}\gamma \mu E_0 \alpha}A. \label{ccoeff}

217: \end{equation}

218: The physical solution corresponds to the + sign in Eq.

219: (\ref{ccoeff}). It can be easily verified that the solution given by

220: Eqs. (\ref{s1})-(\ref{s2}), (\ref{ccoeff}) corresponds to $j_y=0$.

221: Substituting Eq. (\ref{ccoeff}) into Eq. (\ref{eq123}) we finally

222: get

223: \begin{equation}

224: E_y=-\frac{D}{\mu}\alpha-\frac{-1+\sqrt{1+\left( 2\tau_{sf}\gamma

225: \mu E_0 \alpha \right)^2}}{2\tau_{sf} \mu \alpha}. \label{Ey}

226: \end{equation}

227: The second term on the RHS of Eq. (\ref{Ey}) is the electric field

228: needed to compensate the transverse current arising due to the spin

229: Hall effect. If we now assume the sample has a finite (but large)

230: width $L$, then, this term can be interpreted as due to charge

231: accumulation near the edges, as in the ordinary Hall effect.

232: Therefore, the Hall voltage can be written as

233: 

234: \begin{eqnarray}

235: V_H\simeq L \frac{-1+\sqrt{1+\left( 2\tau_{sf}\gamma \mu E_0

236: \alpha \right)^2}}{2\tau_{sf} \mu \alpha}\approx \nonumber \\

237: \approx \left\{

238: \begin{array}{cc}

239: L\tau_{sf} \mu \alpha\gamma^2E_0^2,&  2\tau_{sf}\gamma \mu E_0

240: \alpha \ll 1 \\  \\ L\gamma E_0,& 2\tau_{sf}\gamma \mu E_0 \alpha

241: \gg 1

242: \end{array}

243: \right. . \label{VH}

244: \end{eqnarray}

245: From this equation we see that the Hall voltage is quadratic in

246: $E_0$ for small values of the parameter $2\tau_{sf}\gamma \mu E_0

247: \alpha$, and linear in $E_0$ for large values of this parameter. In

248: fact, the quadratic dependence is quite unusual, since in the

249: ordinary Hall effect the Hall voltage is linear in the longitudinal

250: current. The reason for this unusual dependence can be understood as

251: follows. The charge current in the $y$ direction, determined by the

252: difference of spin-up and spin-down currents, has a component

253: (related to the last term in Eq. (\ref{currentEq})) proportional to

254: the spin density imbalance $P$ times $\gamma E_0$. At small values

255: of $2\tau_{sf}\gamma \mu E_0 \alpha$, the spin density imbalance is

256: proportional to $\gamma E_0$  itself. Therefore, the charge current

257: and Hall voltage are quadratic in $E_0$. At large values of

258: $2\tau_{sf}\gamma \mu E_0 \alpha$, the spin density imbalance

259: saturates and the current dependence on $E_0$ becomes linear.

260: Another difference with respect to the ordinary Hall effect is that

261: the polarity of the Hall voltage in the spin Hall effect is fixed by

262: the geometry of the structure, and does not depend on the direction

263: of the longitudinal current.

264: 

265: Let us now estimate the magnitude of $2\tau_{sf}\gamma \mu E_0

266: \alpha$. Taking parameters related to experiments on GaAs

267: ($\tau_{sf}=10$ns, $\gamma=10^{-3}$(Ref.~\onlinecite{Rashba}),

268: $\mu=8500$cm$^2$/(Vs), $E_0=100$V/cm, $\alpha=2/L$, $L=100\mu$m), we

269: find $2\tau_{sf}\gamma \mu E_0 \alpha=3.4\cdot10^{-3}$. Therefore,

270: in experiments with GaAs, most likely, a quadratic Hall voltage

271: dependence on the longitudinal electric field can be observed.

272: 

273: \begin{figure}[t]

274: \includegraphics[angle=270,width=8.5cm]{fig2a}

275: \includegraphics[angle=270,width=8.5cm]{fig2b}

276: \caption{\label{fig2}(Color online) Distributions of the electron

277: density $n(y)$ and spin density imbalance

278: $P(y)=n_{\uparrow}-n_{\downarrow}$ for a step (a) and exponential

279: (b) background density profiles. The plots presented in the paper

280: were obtained using the parameter values

281:  $\mu=8500$cm$^2$/(Vs), $D=55$cm$^2$/s, $\varepsilon=12.4$,

282:  $\tau_{sf}=10$ns, $\gamma=10^{-3}$, $E_0=100$V/cm and the background density

283:  profiles: (a) $N=10^{16}(1+\theta(y-L/2))$cm$^{-3}$ and (b)

284:  $N=10^{16}\textnormal{exp}(2y/L)$cm$^{-3}$ , where $\theta (..)$ is the

285:  step function, and $L=100\mu $m is the sample width.}

286: \end{figure}

287: 

288: {\it Numerical solution --} Equations (\ref{puaeq})-(\ref{Peq}) can

289: be solved numerically for any reasonable form of $N(y)$. We choose

290: for their simplicity (and possibility to be realized in practice) a

291: step profile and an exponential profile. We solve these equations

292: iteratively, starting with the electron density $n(y)$ close to

293: $N(y)$ and $P(y)$ close to zero and recalculating $E_y(y)$ at each

294: time step.~\cite{prec1} Once the steady-state solution is obtained,

295: the Hall voltage as a function of $E_0$ is calculated as a change of

296: the electrostatic potential across the sample.

297: 

298: \begin{figure}[t]

299: \includegraphics[angle=270,width=8.5cm]{fig3}

300: \caption{\label{fig3}(Color online) Variations in the transverse

301: charge density induced by the longitudinal current. Here, $\delta

302: n=n(E_0=100\textnormal{V/cm})-n(E_0=0)$. The curve for the

303: exponential profile has been shifted vertically by $3\cdot

304: 10^{10}$cm$^{-3}$ for clarity. The dashed lines corresponding to

305: $\delta n=0$ are there to guide the eye.}

306: \end{figure}

307: 

308: Fig. \ref{fig2} shows distributions of the charge density and spin

309: density imbalance in systems with a step (panel (a) of Fig.

310: \ref{fig2}) and exponential (panel (b) of Fig. \ref{fig2})

311: background densities. The values of parameters used for these

312: particular simulations were selected to be close to experimental

313: conditions reported in Ref. \onlinecite{Kato}. However, we have

314: tested the robustness of our predictions by solving Eqs.

315: (\ref{puaeq})-(\ref{Peq}) for different values of parameters, and

316: found that the predicted Hall voltage should be measurable under a

317: wide range of experimental parameters. Quite generally, the

318: self-consistent charge density $n(y)$ is very close to the

319: background density $N(y)$. Small deviations of $n(y)$ from $N(y)$

320: can be observed in regions with strong gradients of $N(y)$. In

321: particular, we can notice that the step profile of electron

322: density in Fig. \ref{fig2}(a) is smoothed out. Such a charge

323: redistribution is related to the diffusion term in Eq. (\ref{CC}).

324: The charge diffusion leads to the formation of a built-in electric

325: field that equilibrates the charge diffusion.

326: 

327: We also find that the induced spin density imbalance $P$ in systems

328: with inhomogeneous electron densities shows some new features, in

329: addition to the well-known spin accumulation near the edges. For

330: instance, in Fig. \ref{fig2}(a), $P$ has an additional peak around

331: $y=50\mu$m. In Fig. \ref{fig2}(b), $P$ is almost constant in the

332: central region of the sample. In both cases, the physics of non-zero

333: spin density imbalance is the same: the spin-current incoming from

334: the right is stronger than the spin-current incoming from the left.

335: We note that the integral spin density imbalance is always zero.

336: 

337: At $E_0=0$, the system is spin-unpolarized and there is no Hall

338: voltage. When the longitudinal current is switched on, the electron

339: charge redistributes, and the associated Hall voltage appears. The

340: change of electron density due to the spin Hall effect is presented

341: in Fig. \ref{fig3}. The first interesting observation is that there

342: is a strong charge accumulation near the edges followed by a charge

343: depletion region. Another observation is that the total electron

344: density in the left region of the samples ($y<50\mu$m) has increased

345: and, correspondingly, the total electron density in the right region

346: has decreased. This change of the electron distribution can be seen

347: in Fig. \ref{fig3}. Therefore, the left part of the samples is

348: charged negatively and the right part is charged positively, as

349: schematically shown in Fig. \ref{fig1}.

350: 

351: The mechanism of formation of sharp peaks of charge accumulation

352: near the edges is similar to the mechanism of formation of Landauer

353: resistivity dipoles \cite{Landauer}. From the point of view of spin

354: currents, the sample edges act as obstacles which block the current

355: flow, and lead to charge accumulation. The adjacent regions with the

356: depleted electron density can be interpreted as screening clouds. We

357: stress that this Landauer-type dipoles of charge accumulation are

358: quite general for spin Hall systems, and should thus be present also

359: in traditionally studied structures with a constant density profile.

360: 

361: \begin{figure}[t]

362: \includegraphics[angle=270,width=8.5cm]{fig4}

363: \caption{\label{fig4}(Color online) Hall voltage as a function of

364: the longitudinal electric field $E_0$.}

365: \end{figure}

366: 

367: We finally plot in Fig. \ref{fig4} the change of the electrostatic

368: potential across the sample as a function of longitudinal electric

369: field. The Hall voltage, for both density profiles, has a

370: dependence on $E_0$ which is very close to the exponential

371: dependence we have predicted analytically in Eq.~\ref{VH} for

372: small values of $2\tau_{sf}\gamma \mu E_0 \alpha$. The fact that

373: this exponential dependence appears also in the step profile,

374: hints at a possible ``general'' property of the Hall voltage in

375: spin Hall systems with inhomogeneous densities. We emphasize that

376: a Hall voltage should also appear in spin Hall systems with a

377: homogeneous electron density, but inhomogeneous $\gamma$. This

378: corresponds to the case in which the spin-orbit coupling is

379: dependent on space\cite{serra,nitta}.

380: 

381: In conclusion, we have shown that a Hall voltage would appear in

382: spin Hall systems with inhomogeneous electron density in the

383: direction perpendicular to main current flow. The striking result is

384: that this Hall voltage is generally quadratic in the longitudinal

385: electric field, unlike the ordinary Hall voltage which is linear in

386: the same field. These results can be easily verified experimentally,

387: and would simplify tremendously the measurement of the spin Hall

388: effect by allowing an electrical measurement of the latter in

389: non-magnetic systems, and without injection of spin-polarized

390: electrons.

391: 

392: This work is partly supported by the NSF Grant No. DMR-0133075.

393: 

394: \begin{thebibliography}{}

395: 

396: \bibitem{natureSHC} J. Inoue and H. Ohno, Science {\bf 309}, 2004

397: (2005).

398: 

399: \bibitem{r0} M. I. D'yakonov and V. I. Perel, Phys. Lett. A {\bf 35}, 459

400: (1971).

401: 

402: \bibitem{r1} J. E. Hirsch, Phys. Rev. Lett. {\bf 83}, 1834 (1999).

403: 

404: \bibitem{r2} S. Murakami, N. Nagaosa, S.-C. Zhang, Science {\bf 301}, 1348

405: (2003).

406: 

407: \bibitem{r3} J. Sinova, D. Culcer, Q. Niu, N. A. Sinitsyn, T. Jungwirth,

408: and A. H. MacDonald, Phys. Rev. Lett. {\bf 92}, 126603 (2004).

409: 

410: \bibitem{Rashba} H.-A. Engel, B. I. Halperin, and E. I. Rashba

411: Phys. Rev. Lett. {\bf 95}, 166605 (2005).

412: 

413: \bibitem{Rashba1} H.-A. Engel, E. I. Rashba, and B. I. Halperin

414: Phys. Rev. Lett. {\bf 98}, 036602 (2007).

415: 

416: \bibitem{r4} J. Inoue, G. E. W. Bauer, and L. W. Molenkamp, Phys. Rev. B {\bf 70}, 041303(R) (2004).

417: 

418: \bibitem{r5} B. A. Bernevig and S.-C. Zhang, Phys. Rev. Lett. {\bf 95}, 016801 (2005).

419: 

420: \bibitem{r6} M. W. Wu and J. Zhou, Phys. Rev. B {\bf 72}, 115333

421: (2005).

422: 

423: \bibitem{r7} E. M. Hankiewicz, G. Vignale, and M. E. Flatt\'e, Phys. Rev. Lett. {\bf 97}, 266601

424: (2006).

425: 

426: \bibitem{r8} B. K. Nikoli\'c, S. Souma, L. P. Z\^arbo, and J.

427: Sinova, Phys. Rev. Lett. {\bf 95}, 046601 (2005).

428: 

429: \bibitem{r9} J. I. Inoue, G. E. W. Bauer, L. W. Molenkamp, Phys. Rev. B {\bf 70}, 041303(R)

430: (2004).

431: 

432: \bibitem{r10} A. G. Mal'shukov and C. S. Chu, Phys. Rev. Lett. {\bf 97}, 076601

433: (2006).

434: 

435: \bibitem{Kato} Y. K. Kato, R. C. Myers, A. C. Gossard, D. D. Awschalom,

436: Science {\bf 306}, 1910 (2004).

437: 

438: \bibitem{Wunderlich} J. Wunderlich, B. Kaestner, J. Sinova, and T.

439: Jungwirth, Phys. Rev. Lett. {\bf 94}, 047204 (2005).

440: 

441: \bibitem{Valenzuela} S. O. Valenzuela and M. Tinkham, Nature {\bf 442}, 176

442: (2006).

443: 

444: \bibitem{Landauer} R. Landauer, IBM J. Res. Develop. {\bf 1}, 223

445: (1957).

446: 

447: \bibitem{YP} Y. V. Pershin and V. Privman, Phys. Rev. Lett. {\bf 90}, 256602

448: (2003).

449: 

450: \bibitem{flatte} Z. G. Yu and M. E. Flatt\'e, Phys. Rev. B {\bf 66}, 201202 (2002).

451: 

452: \bibitem{PDV} Y. V. Pershin and M. Di Ventra, cond-mat/0701678.

453: 

454: \bibitem{prec} This is a good approximation for the range of

455: parameters considered in this work.

456: 

457: \bibitem{prec1} We have employed the Scharfetter-Gummel discretization scheme

458: [D. L. Scharfetter and H. K. Gummel,  IEEE. Trans. Electron.

459: Devices, {\bf ED-16}, 64 (1969)] to solve both Eqs. (\ref{CC}) and

460: (\ref{Peq}) numerically.

461: 

462: \bibitem{serra} M. Val\'{\i}n-Rodr\'{\i}guez, A. Puente, and L. Serra

463: Phys. Rev. B {\bf 69}, 153308 (2004).

464: 

465: \bibitem{nitta} J. Nitta, T. Akazaki, and H. Takayanagi, Phys. Rev. Lett. {\bf 78}, 1335 (1997).

466: 

467: \end{thebibliography}

468: 

469: \end{document}

470: