1: %\documentstyle[version2,aps]{revtex}

2: %\documentstyle[12pt]{article}

3: %\documentstyle[prl,aps,preprint,tighten]{revtex}

4: %\documentstyle[prl,aps,preprint]{revtex}

5: \documentstyle[prl,twocolumn,aps]{revtex}

6: %\usepackage{graphicx}

7: \input psfig

8: \begin{document}

9: \draft

10: 

11: %----------------------User's Commands----------------------------

12:  \newcommand{\mytitle}[1]{

13:  \twocolumn[\hsize\textwidth\columnwidth\hsize

14:  \csname@twocolumnfalse\endcsname #1 \vspace{1mm}]}

15: %------------------------------------------------------------------------

16: 

17: \mytitle{

18: \title{Nonperturbative analysis of coupled quantum dots in a phonon bath}

19: 

20: \author{Markus Keil$^1$ and Herbert Schoeller$^1$}

21: 

22: \address{

23: $^1$ Institut f\"ur Theoretische Physik, Rheinisch-Westf\"alische

24: Technische Hochschule Aachen, 52056 Aachen, Germany

25: }

26: 

27: \date{\today}

28: 

29: \maketitle

30: 

31: \begin{abstract}

32: Transport through coupled quantum dots in a phonon bath is studied using

33: the recently developed real-time renormalization-group

34: method. Thereby, the problem can be treated beyond perturbation theory

35: regarding the complete interaction. A

36: reliable solution for the stationary tunnel current is obtained for

37: the case of moderately strong couplings of

38: the dots to the leads and to the phonon bath. Any other parameter is

39: arbitrary, and the

40: complete electron-phonon interaction is taken into

41: account. Experimental results are

42: quantitatively reproduced by taking into account a finite extension of

43: the wavefunctions within the dots. Its dependence on the energy

44: difference between the dots is derived.

45: \end{abstract}

46: \pacs{73.23.Hk,71.38.-k,05.10.Cc}

47: }

48: 

49: {\it Introduction}.

50: Today quantum dot systems allow a detailed study of many physical

51: phenomena, like Coulomb blockade~\cite{raikh,tarucha}, Kondo

52: effects~\cite{cronenwett,wiel,sasaki}, interference

53: effects~\cite{wiel}. As in these structures quantum states can be

54: manipulated, they also may have an application in future quantum

55: gates~\cite{loss}. 

56: Quantum dot systems can typically be characterized by only a few parameters,

57: which are experimentally controllable. Theoretically, these systems

58: may be described by basic models, which capture the essential

59: physics and can be investigated using standard methods of many-particle

60: theory. However, if one deals with out-of-equilibrium situations and

61: strong coupling, a

62: theoretical analysis becomes difficult.

63: 

64: Such an out-of-equilibrium problem was recently studied, when the

65: stationary tunnel current through a double quantum dot in a

66: phonon bath was measured~\cite{fujisawa}. There the influence of the 

67: phonon environment was examined at low temperature. The thermal energy

68: of the environment

69: is always a source of unwanted transitions in quantum dot

70: devices. Even at zero temperature spontaneous emission of phonons

71: gives rise to inelastic transitions, i.e. they occur between dot

72: states of nonequal energy. In the experiment in Ref.~\cite{fujisawa} the

73: inelastic contribution to the tunnel current through the double dot

74: was studied. A first

75: theoretical interpretation of the experimental results focused on the

76: interaction

77: of the dots with the phonons, which is analogous to the spin-boson

78: model~\cite{brandes}. They accounted for the coupling to the leads

79: perturbatively and studied the electron-phonon problem using an approximation,

80: which corresponds to the

81: noninteracting blip approximation (NIBA) for the spin-boson

82: model~\cite{leggett-etal}. Thereby a better understanding of the shape of the

83: inelastic current

84: spectrum was achieved. It was shown that the interference

85: of phonons interacting with the electron densities at the two dots

86: leads to an oscillating structure in the current spectrum.

87: However, a quantitative comparison

88: with the experiment has not yet been possible. Especially, the

89: unexpectedly large

90: inelastic current of the experiment could not be explained.

91: 

92: In this paper we present a reliable solution for the

93: stationary state of the double dot system. Since we

94: deal with a nonequilibrium situation, where both the coupling to the

95: phonon reservoir and the coupling to the leads have to be considered

96: as strong, we use the

97: recently developed real-time

98: renormalization-group (RTRG) method. It has successfully been applied

99: to both equilibrium~\cite{koenig-hs,mk-hs_pol} and nonequilibrium

100: problems~\cite{hs-koenig,mk-hs_sbm}, including the spin-boson

101: model~\cite{mk-hs_sbm,mk-hs_sbmproc}.

102: Applying this approach to the double dot system both the electron and

103: the phonon reservoirs are integrated out in a RG procedure,

104: i.e. beyond perturbation theory. Thereby, the external voltage is

105: accounted for properly, and the level broadening induced by

106: the coupling to the leads is included in this method. Thus, we

107: can also consider the case,

108: where the energy difference between the dots is of the same order as

109: the external voltage. Furthermore, we do not have to include an

110: additional cutoff parameter, which simulates the level broadening

111: generated by the leads. Moreover,

112: this formalism may treat any form of dot-phonon interaction. Therefore, we

113: are able to account for the full electron-phonon interaction, i.e.

114: including interaction terms, which involve a tunneling between the dots

115: (``offdiagonal interaction terms'').

116: In the quantitative analysis

117: the offdiagonal interaction terms lead to a strong

118: dependence of the current on the extension of the

119: wavefunctions within one dot. We find that the variation of this extension

120: with the energy difference $\epsilon$ between the dot levels has to be

121: accounted for. By fitting the result for the current with the

122: experimental data, we obtain the width of the electron density within one

123: dot as a function of $\epsilon$.

124: Furthermore, for the physically relevant situation, where the

125: coupling of the dots to the leads and to the phonons is only

126: moderately large, the RTRG approach does not deal with any

127: parameter restriction. This is in contrast to the NIBA, which is valid

128: only for sufficiently high temperature.

129: 

130: {\it Model}.

131: Our model consists of two coupled quantum dots ($l$ and $r$,

132: respectively). Each dot is coupled to an electron reservoir with the

133: chemical potentials $\mu_l$ and $\mu_r$, see Fig.~\ref{ddfig}. We consider

134: the case realized

135: in the experiment~\cite{fujisawa}, where

136: the external voltage $V=\mu_l-\mu_r$ is much smaller than the Coulomb

137: charging energy $U$. Thus, due to Coulomb blockade the double dot

138: cannot be charged with more than one additional electron. In the

139: experiment~\cite{fujisawa} a strong magnetic field was applied

140: perpendicular to the dots. Thus, we assume spin polarization here and

141: omit the spin index. 

142: We denote the many-particle ground states, where an additional electron is in

143: the left (right) dot, by $|l\rangle$ ($|r\rangle$) and neglect any

144: excited states.

145: Therefore, together with the uncharged ground state $|0\rangle$,

146: there are only three possible states of the double dot. 

147: The total Hamiltonian $\bar H$ for the system can be written as a sum

148: of the dot

149: Hamiltonian, the contributions of the electron reservoirs and the

150: phonon bath, and the interaction parts stemming from the coupling to

151: the leads and the electron-phonon interaction:

152: \begin{equation}

153: \label{hsumeq}

154: \bar H = H_d + H_{res} + H_{ph} + H_{e-res} + H_{e-ph}\,.

155: \end{equation}

156: The dot Hamiltonian $H_d$ reads

157: \begin{equation}

158: \label{hdeq}

159: H_d = \epsilon_l |l\rangle\langle l| + \epsilon_r |r\rangle\langle r|

160: + T_c \left(|l\rangle\langle r| + |r\rangle\langle l|\right)\,,

161: \end{equation}

162: where $\epsilon_l$ ($\epsilon_r$) are the ground state energies of

163: $|l\rangle$ ($|r\rangle$) and the coupling between the dots is

164: described by the tunnel matrix element $T_c$.

165: The reservoir contributions are given by

166: \begin{eqnarray}

167: \label{hreq}

168: H_{res} &=& \sum_k \epsilon_k c^{\dagger}_kc_k + \sum_k \epsilon_k

169: d^{\dagger}_kd_k\,,\\

170: \label{hpeq}

171: H_{ph} &=& \sum_q \omega_q a^{\dagger}_qa_q\,.

172: \end{eqnarray}

173: Here, the operators $c^{\dagger}_k$ ($c_k$) create (annihilate) an

174: electron with the energy $\epsilon_k$ in the left lead, whereas the

175: creation (annihilation) operators $d^{\dagger}_k$ ($d_k$) refer to the

176: right electron reservoir. Analogously, $a^{\dagger}_q$ ($a_q$) create

177: (annihilate) a phonon with the wavevector $\vec q$ and the frequency

178: $\omega_q$. Here and

179: throughout the paper we set $\hbar=1$. 

180: The double dot is coupled to the external leads by the parameters $V_k$

181: and $W_k$:

182: \begin{eqnarray}

183: H_{e-res} &=& \sum_k \left(V_k c_k|l\rangle\langle 0| + V^*_k

184: |0\rangle\langle l|c^{\dagger}_k\right)\nonumber\\

185: && + \sum_k \left(W_k d_k|r\rangle\langle

186: 0| + W^*_k |0\rangle\langle r|d^{\dagger}_k\right)\,.

187: \label{hereq}

188: \end{eqnarray}

189: The electron-phonon interaction consists of a diagonal part, which is

190: characterized by the coupling constants $\alpha_q$ and $\beta_q$, and

191: an offdiagonal contribution with the parameter $\gamma_q$:

192: \begin{eqnarray}

193: H_{e-ph} &=& \sum_q (\alpha_q|l\rangle\langle l| +

194: \beta_q|r\rangle\langle r|)(a^{\dagger}_q + a_{-q})\nonumber\\

195: && + \sum_q \gamma_q(|l\rangle\langle r| +

196: |r\rangle\langle l|)(a^{\dagger}_q + a_{-q})\,.

197: \label{hepeq}

198: \end{eqnarray}

199: The above interaction coefficients are given by~\cite{brandes}

200: \begin{eqnarray}

201: \label{alphaqeq}

202: \alpha_q &=& \lambda_q \langle l|e^{i \vec q \vec x}|l\rangle\,,\\

203: \label{betaqeq}

204: \beta_q &=& \lambda_q \langle r|e^{i \vec q \vec x}|r\rangle\,,\\

205: \label{gammaqeq}

206: \gamma_q &=& \lambda_q \langle l|e^{i \vec q \vec x}|r\rangle\,,

207: \end{eqnarray}

208: where $\lambda_q$ is the matrix element for the interaction of 2DEG

209: electrons and phonons. The phonons are assumed to be three-dimensional

210: acoustical phonons~\cite{fujisawa}. It then follows for the

211: interaction~\cite{brandes}

212: \begin{equation}

213: \label{lambdaqeq}

214: |\lambda_q|^2 = g \frac{\pi^2 c_s^2}{V|\vec q|}\,,

215: \end{equation}

216: and the dispersion reads

217: \begin{equation}

218: \label{omegaqeq}

219: \omega_q = c_s |\vec q|\,.

220: \end{equation}

221: Here, we introduced $c_s$ as the speed of sound in the medium, $V$

222: as the volume of the crystal, and the dimensionless coupling constant

223: $g$~\cite{brandes}.

224: For the evaluation of Eqs.~(\ref{alphaqeq}) - (\ref{gammaqeq}) we

225: model the electron densities $\rho_l(\vec x)$ ($\rho_r(\vec x)$) within one

226: dot by Gaussians, which are peaked around the dot positions $x_l$

227: ($\vec x_r = \vec

228: x_l + \vec d$) with a width $|\Delta\vec x|=\sqrt{3/2}\,\sigma$:

229: \begin{equation}

230: \label{rholreq}

231: \rho_{l(r)}(\vec x) = \left(\frac{1}{\pi\sigma^2}\right)^{3/2}

232: e^{-\frac{(\vec x - \vec x_{l(r)})^2}{\sigma^2}}\,.

233: \end{equation}

234: The finite width $\sigma$ leads to a high-energy cutoff $D=c_s/\sigma$ for the

235: coefficients $\alpha_q$, $\beta_q$ and $\gamma_q$. We include this

236: cutoff in an exponential form, so that we end up with the following

237: interaction coefficients:

238: \begin{eqnarray}

239: \label{alphaq1eq}

240: \alpha_q &=& \lambda_q e^{i \vec q \vec x_l} e^{-\frac{c_s|\vec q|}{2D}}\,,\\

241: \label{betaq1eq}

242: \beta_q &=& \lambda_q e^{i \vec q \vec x_r} e^{-\frac{c_s|\vec q|}{2D}}\,,\\

243: \label{gammaq1eq}

244: \gamma_q &=& \lambda_q e^{i \vec q (\frac{\vec x_l + \vec x_r}{2})}

245: e^{-\frac{|\vec d|D}{2c_s}} e^{-\frac{c_s|\vec q|}{2D}}\,.

246: \end{eqnarray}

247: A simple form of the Hamiltonian, which shows the analogy with the

248: spin-boson model, is obtained by shifting the bosonic field

249: operators. One introduces the unitary transformation

250: \begin{equation}

251: \label{ueq}

252: U =

253: \exp\left[\sum_q\left(\frac{\alpha_q+\beta_q}{2\omega_q}a^{\dagger}_q

254: - \frac{\alpha^*_q+\beta^*_q}{2\omega_q}a_q\right)\right]\,,

255: \end{equation}

256: so that

257: \begin{equation}

258: \label{bqeq}

259: Ua_qU^{\dagger} = a_q - \frac{\alpha_q + \beta_q}{2\omega_q}\,.

260: \end{equation}

261: Thus, our final Hamiltonian $H=U\bar HU^{\dagger}$ reads

262: \begin{eqnarray}

263: \label{hsumfinaleq}

264: H &=& H_0 + H_B + H_V\,,\\

265: H_0 &=& \frac{\epsilon}{2} \left(|l\rangle\langle l| -

266: |r\rangle\langle r|\right)\nonumber\\

267: && + T_c^{\rm eff}

268: \left(|l\rangle\langle r| + |r\rangle\langle l|\right)\nonumber\\

269: && + E

270: \left(|l\rangle\langle l| + |r\rangle\langle r| - |0\rangle\langle 0|\right)

271: \label{h0finaleq}

272: \,,\\

273: \label{hbfinaleq}

274: H_B &=& \sum_k \epsilon_k c^{\dagger}_kc_k + \sum_k \epsilon_k

275: d^{\dagger}_kd_k + \sum_q \omega_q a^{\dagger}_qa_q\,,\\

276: \label{hvfinaleq}

277: H_V &=& \sum_{\mu}:g_{\mu}j_{\mu}:\,,

278: \end{eqnarray}

279: where we have used Eqs.~(\ref{omegaqeq}) and (\ref{alphaq1eq}) -

280: (\ref{gammaq1eq}). Furthermore, for simplicity we have

281: set $(\epsilon_l+\epsilon_r)/2=0$ and have introduced the parameters

282: \begin{eqnarray}

283: \label{epseq}

284: \epsilon &=& \epsilon_l-\epsilon_r\,,\\

285: \label{Tceq}

286: T_c^{\rm eff} &=& T_c -

287: 2g\omega_d e^{-\frac{D}{2\omega_d}}\arctan\frac{D}{2\omega_d}\,,\\

288: \label{Eeq}

289: E &=& -\frac{g}{4} \left(D +

290: \omega_d\arctan\frac{D}{\omega_d}\right)

291: \end{eqnarray}

292: and

293: \begin{equation}

294: \label{omdeq}

295: \omega_d = \frac{c_s}{|\vec d|}\,.

296: \end{equation}

297: Thus, the tunnel

298: amplitude $T_c$ has to be replaced by a smaller effective $T_c^{\rm eff}$,

299: which is due to the offdiagonal electron-phonon interaction. One

300: already recognizes that the reduction of $T_c$ strongly depends on the

301: width of the electron

302: densities $\sigma=|\vec d|\omega_d/D$.

303: Finally, in view of the RTRG method, we have written the interaction part $H_V$

304: as normal ordered products of local (dot) operators $g_{\mu}$ and environmental

305: operators $j_{\mu}$.

306: They are defined by

307: \begin{eqnarray}

308: \label{gb1eq}

309: g_{b_1} &=& \frac{1}{2}\left(|l\rangle\langle l| - |r\rangle\langle

310: r|\right)\,,\\

311: \label{jb1eq}

312: j_{b_1} &=& \sum_q \left(\alpha_q - \beta_q\right) \left(a^{\dagger}_q +

313: a_{-q}\right)\,,\\

314: \label{gb2eq}

315: g_{b_2} &=& \left(|l\rangle\langle r| + |r\rangle\langle

316: l|\right)\,,\\

317: \label{jb2eq}

318: j_{b_2} &=& \sum_q \gamma_q \left(a^{\dagger}_q +

319: a_{-q}\right)\,,\\

320: \label{gb3eq}

321: g_{b_3} &=& -\frac{1}{2} |0\rangle\langle 0|\,,\\

322: \label{jb3eq}

323: j_{b_3} &=& \sum_q \left(\alpha_q + \beta_q\right) \left(a^{\dagger}_q +

324: a_{-q}\right)\,,\\

325: \label{gjleq}

326: g_{+l} &=& g_{-l}^{\dagger} = |l\rangle\langle 0| \quad,\qquad j_{+l}

327: = j_{-l}^{\dagger} = \sum_k V_k c_k\,,\\

328: \label{gjreq}

329: g_{+r} &=& g_{-r}^{\dagger} = |r\rangle\langle 0| \quad,\qquad j_{+r}

330: = j_{-r}^{\dagger} = \sum_k W_k d_k\,.\\

331: \end{eqnarray}

332: Therefore, the interaction index $\mu$ runs over the bosonic indices

333: $b_1,b_2,b_3$ and the fermionic ones $+l,-l,+r,-r$.

334: From Eqs.~(\ref{hsumfinaleq}) - (\ref{hvfinaleq}) the spin-boson model is

335: recovered by omitting

336: the electron reservoirs and the interaction with them, excluding

337: the state $|0\rangle$ and accounting only for the bosonic interaction

338: $b_1$. The latter corresponds to neglecting the offdiagonal

339: electron-phonon ($b_2$) interaction.

340: 

341: {\it RTRG approach}.

342: The RTRG approach is based on the formally exact kinetic equations

343: \begin{eqnarray}

344: \label{iteq}

345: \langle I\rangle (t) &=& {\rm Tr}_0

346: \left[\int_0^tdt'\, \Sigma_I(t-t')p(t')\right]\,,\\

347: \dot{p}(t) + iL_0p(t) &=& \int_0^tdt'\, \Sigma(t-t')p(t')\,,

348: \label{pteq}

349: \end{eqnarray}

350: where $\langle I\rangle(t)$ denotes the time-dependent expectation

351: value of the current through the double dot, and $p(t)$ is the

352: time-dependent reduced density matrix, which is the trace over the bath

353: degrees of freedom ${\rm Tr}_B$ of the full density matrix. In contrast, ${\rm

354: Tr}_0$ denotes the trace over the local (dot) degrees of freedom. The

355: bath degrees of freedom enter the equations via the

356: integral kernels $\Sigma$ and $\Sigma_I$, which are defined by sums

357: over irreducible diagrams, see Ref.~\cite{hs} for details.

358: Introducing the Laplace transforms $f(z)=\int_0^{\infty}dt\,

359: e^{izt}f(t)$

360: of the time-dependent functions $f(t)$ and using the identity

361: $\lim_{t\rightarrow\infty}f(t)=-i\lim_{z\rightarrow 0}zf(z)$ leads to

362: the following expression for the stationary tunnel

363: current $I_{st}$:

364: \begin{equation}

365: \label{isteq}

366: I_{st} = {\rm Tr}_0\left[\Sigma_I(z=0)p_{st}\right]\,,

367: \end{equation}

368: where the stationary reduced density matrix $p_{st}$ is determined by

369: \begin{equation}

370: \label{psteq}

371: \left(L_0 + i\Sigma(z=0)\right)p_{st} = 0\,.

372: \end{equation}

373: In the following we outline only the main steps of the (i) derivation

374: of Eqs.~(\ref{iteq}) and (\ref{pteq}), and of the (ii) RTRG technique, by

375: which $\Sigma$ and

376: $\Sigma_I$ are calculated.

377: 

378: (i) The quantity of interest is the current through the double dot. It is

379: given by the expectation value of the operator

380: \begin{equation}

381: \label{idefeq}

382: I = ie \sum_k \left(V^*_k|0\rangle\langle l|c^{\dagger}_k -

383: V_kc_k|l\rangle\langle 0|\right)

384: \end{equation}

385: with $e$ being the elementary charge.

386: For the calculation of $\langle I\rangle(t)$ we introduce the

387: Liouvillian $L=[H,\cdot]_-$ and the

388: superoperator $A_I=[\frac{i}{2}I,\cdot]_+$, where

389: $[\cdot,\cdot]_{-(+)}$ denotes the (anti)commutator. The expectation

390: value of the current can then be written as

391: \begin{equation}

392: \label{it1eq}

393: \langle I\rangle (t) = -i {\rm Tr} \left[A_I e^{-iLt} p(0)\rho_B^{eq}\right]\,.

394: \end{equation}

395: Here, we assumed a factorized density matrix at $t=0$, where $p(0)$ is

396: the initial dot density matrix and $\rho_B^{eq}$ is the equilibrium

397: distribution of the electron reservoirs and the phonon bath.

398: To evaluate the above expression the propagator $\exp{(-iLt)}$ is

399: expanded in the interaction 

400: part $L_V=[H_V,\cdot]$. According to Eq.~(\ref{hvfinaleq}) this

401: can be written as

402: \begin{equation}

403: \label{lvlepeq}

404: L_V = \sum_{p\mu}:G^p_{\mu}J^p_{\mu}:\,,

405: \end{equation}

406: where the index $p=\pm$ denotes, whether the interaction takes place on

407: the forward or backward propagator, i.e.

408: \begin{eqnarray}

409: \label{gpeq}

410: G^+_{\mu} &=& g_{\mu}\,\cdot \quad ,\qquad G^-_{\mu} = \cdot\,

411: \left(-g_{\mu}\right)\,,\\

412: \label{jpeq}

413: J^+_{\mu} &=& j_{\mu}\,\cdot \quad ,\qquad J^-_{\mu} = \cdot\,

414: j_{\mu}\,.

415: \end{eqnarray}

416: Correspondingly, we write $A_I \!=\! \sum_p\!\! :\!\!(A^p_{I+l}J^p_{+l} +

417: A^p_{I-l}J^p_{-l})\!\!:$ with

418: \begin{eqnarray}

419: \label{ap+eq}

420: A^+_{I+l} &=& \frac{e}{2}|l\rangle\langle 0| \,\cdot \quad ,\qquad

421: A^-_{I+l} = \cdot\, \frac{e}{2}|l\rangle\langle 0|\,,\\

422: \label{ap-eq}

423: A^+_{I-l} &=& -\frac{e}{2}|0\rangle\langle l| \,\cdot \quad ,\qquad

424: A^-_{I-l} = \cdot\, \left(-\frac{e}{2}|0\rangle\langle l|\right)\,.

425: \end{eqnarray}

426: The trace over the bath degrees

427: of freedom can then be 

428: performed by application of Wick's theorem.

429: In this way one obtains a 

430: series of terms, where vertices $G^p_{\mu}$ and $A^p_{I\pm l}$ of the local

431: system are connected

432: by pair contractions $\gamma^{pp'}_{\mu\mu'}(t)={\rm

433: Tr}_B\left[J^p_{\mu}J^{p'}_{\mu'}\rho_B^{eq}\right]$ of the bath.

434: Denoting the sum over all irreducible diagrams, which contain the

435: current superoperators $A^p_{I\pm l}$, by $\Sigma_I$ then leads to

436: Eq.~(\ref{iteq}), see Fig.~\ref{itfig}.

437: Similarly, Eq.~(\ref{pteq}) is obtained by introducing the object

438: $\Sigma$, which is defined as the sum over

439: all irreducible diagrams involving only the vertices

440: $G^p_{\mu}$.

441: 

442: (ii) The objects $\Sigma(z)$ and $\Sigma_I(z)$ are calculated by a

443: renormalization group procedure. Short time scales of

444: $\gamma^{pp'}_{\mu\mu'}(t)$ are

445: integrated out first 

446: by introducing a short-time cutoff $t_c$ into the correlation

447: function

448: $\gamma^{pp'}_{\mu\mu'}(t)\rightarrow\gamma^{pp'}_{\mu\mu'}(t,t_c)$.

449: In each renormalization-group step, the time scales between $t_c$ and

450: $t_c+dt_c$ are 

451: integrated out, starting from $t_c=0$ and ending at $t_c=\infty$.

452: As a consequence, one generates RG equations for $\Sigma(z)$ ($\Sigma_I(z)$), 

453: $L_0$, $G^p_{\mu}$, and the boundary vertex operators $A^p_{\mu}$

454: ($A^p_{I\mu}$)and $B^p_{\mu}$ 

455: (defined as the rightmost and leftmost vertex of the kernel

456: $\Sigma(z)$ ($\Sigma_I(z)$)), for the explicit form of the RG

457: equations see the Appendix.

458: 

459: Within the scheme of a perturbative RG analysis, the generation of multiple

460: vertex superoperators is neglected here, as in Ref.~\cite{hs-koenig,mk-hs_sbm}.

461: Due to the small realistic value of $g$ ($g=0.05$ for GaAs) the

462: neglecting of double- and

463: higher-order vertex objects is justified and our approach leads to very

464: reliable results, see Ref.~\cite{mk-hs_sbm}, where we solved the spin-boson

465: model for couplings up to $\alpha=g/2\lesssim 0.1..0.2$. 

466: Furthermore, since also the coupling to the leads is treated

467: nonperturbatively, the induced level broadening is accounted for

468: properly. In contrast, in a perturbative analysis this effect has to be

469: accounted for by an additional cutoff parameter.

470: Also note that, in our method there is no restriction regarding the

471: temperature, whereas the NIBA is only valid for

472: the parameter regime $(\Delta_r^2+\epsilon^2)^{1/2}\lesssim

473: T$~\cite{weiss-book}. Here,

474: $\Delta_r=2T_c(2T_c/D)^{\alpha/(1-\alpha)}$ is the renormalized tunnel

475: amplitude of the spin-boson model.

476: Finally, using the RTRG we are also able to account for the offdiagonal

477: electron-phonon interaction, which is important for small

478: $D/\omega_d$, see Eqs.~(\ref{Tceq}), (\ref{r22eq}) - (\ref{s23eq}).

479: 

480: {\it Results}.

481: We solve the set of ordinary differential equations,

482: Eqs.~(\ref{dsigmaeq}) - (\ref{dbeq}), numerically. The stationary tunnel

483: current then follows from Eqs.~(\ref{isteq}) and (\ref{psteq}). Our choice

484: of the parameters corresponds to the experiment, where a GaAs

485: structure was used at the temperature

486: $T=23mK=1.98\mu eV$ with an external voltage $V=140\mu

487: eV$~\cite{fujisawa}. For GaAs we have $c_s=5000m/s$ and

488: $g=0.05$~\cite{bruus}. The distance between the dots is estimated as

489: $d=200\cdot 10^{-9}m$~\cite{brandes}, which leads to $\omega_d=16.5\mu

490: eV$.

491: 

492: The result for $T_c=\Gamma_l=\Gamma_r=1\mu eV$, $D_l=D_r=1meV$ and

493: $D=100\mu eV$ respectively $D=150\mu eV$, which corresponds to the

494: parameters studied in

495: Ref.~\cite{brandes}, is shown in Fig.~\ref{ist_brartrgfig}. The external

496: voltage $V$ gives rise to a finite stationary tunnel current through

497: the double dot. The elastic

498: current can be seen around $\epsilon$ with a width depending on the

499: coupling to the leads $\Gamma_l=\Gamma_r$ and the internal tunnel

500: amplitude $T_c$. There the phonons do not

501: participate in the tunnel process. Due to the

502: coupling to the phonons

503: there is also an inelastic current, where phonons are emitted

504: ($\epsilon>0$) respectively absorbed ($\epsilon<0$) during the tunnel

505: process. For increasing $\Gamma_{l(r)}$ the width of the elastic current grows,

506: while an increased coupling constant $g$ leads to a larger inelastic

507: current. The effect of

508: the finite voltage $V$ can be seen in Fig.~\ref{ist_brartrgfig},

509: where for $\epsilon>V$ the tunnel current drops to zero.

510: Furthermore, Fig.~\ref{ist_brartrgfig} shows that the offdiagonal

511: interaction leads to a larger inelastic

512: current. This

513: effect is

514: increased with decreasing $D$, see also Eqs.~(\ref{r22eq}) -

515: (\ref{s23eq}). We will see below

516: (Fig.~\ref{ist_exprtrgfig}) that, due to Eq.~(\ref{Tceq}), the

517: offdiagonal interaction also

518: suppresses the elastic current. 

519: Eventually, in Fig.~\ref{ist_brartrgfig} one also recognizes the oscillations

520: stemming from

521: the interference of the phonons interacting with the two dots~\cite{brandes}.

522: 

523: Let us now study the current quantitatively in comparison with the experiment.

524: For this it is necessary to choose realistic parameter values for

525: $T_c$, $\Gamma_{l(r)}$, $D$ and

526: $D_{l(r)}$. From changing the bias polarity in the experiments the

527: ratio $\Gamma_r/\Gamma_l\approx 0.5..1$ was found~\cite{fujisawa}. To

528: determine the

529: couplings $T_c$ and $\Gamma_r$, the

530: experimental data are compared with the result of Stoof and

531: Nazarov~\cite{stoof}

532: \begin{equation}

533: \label{stoofeq}

534: I_{st} = \frac{T_c^2\Gamma_r}{T_c^2(2+\Gamma_r/\Gamma_l) + \Gamma_r^2/4 + \epsilon^2}\,,

535: \end{equation}

536: which is valid for no electron-phonon interaction ($g=0$).

537: A good agreement of the elastic current is found for $T_c=0.124\mu

538: eV$ and $\Gamma_l=\Gamma_r=3.5\mu eV$, see Fig.~\ref{ist_expstooffig}.

539: 

540: However, due to the absent electron-phonon interaction the influence

541: of the finite extension of the electron

542: densities within the dots is also disregarded in the Stoof-Nazarov result.

543: In contrast, our method accounts for this extension by the high-energy

544: cutoff $D$. From Eq.~(\ref{Tceq}) we see that for a finite $D$ the

545: tunnel amplitude is effectively reduced. Therefore, the Stoof-Nazarov

546: result underestimates the

547: value of $T_c$.

548: 

549: From the large inelastic current in the experiment one can conclude,

550: that in fact a finite value of $D$ was realized. It turns out that

551: $T_c\approx 0.375$ allows sensible fits.

552: First, in Fig.~\ref{ist_exprtrgfig} our results for

553: $T_c=0.375\mu eV$, $\Gamma_l=\Gamma_r=3.5\mu eV$, $D_l=D_r=1meV$ and

554: $D=70\mu eV$

555: respectively $D=100\mu eV$ are compared with the experiment. One 

556: recognizes that with decreasing $D$ the larger overlap of the dots'

557: wavefunctions leads to a stronger impact of the offdiagonal

558: electron-phonon interaction. The elastic current is suppressed, whereas

559: the inelastic current is increased. The

560: deviations from the experiment show, that there is an $\epsilon$

561: dependence of the width of the

562: electron densities, which we have to account for in order to achieve

563: agreement. In Fig.~\ref{sig_rtrgfig} we show a fit of the width

564: $\sigma=d\omega_d/D$, which is based on the experimental results for

565: $I_{st}$. One

566: recognizes that for larger absolute values of $\epsilon$ the electron

567: densities are

568: more sharply peaked. The asymmetry is due to the finite external

569: voltage $V$. For $\epsilon<0$ the state $|l\rangle$ lies in a deep

570: potential well, thus this energetic separation of the two quantum

571: dot levels and the leads causes a very small overlap of the

572: wavefunctions within the dots. On the other hand for $\epsilon>0$

573: neither dot level

574: lies in a deep potential well, however, an increasing energetic

575: separation $\epsilon$ again leads to more sharply defined electron densities.

576: 

577: In Fig.~\ref{ist_exprtrgfig} one also recognizes that the structure on

578: the emission side ($\epsilon>0$) of the current spectrum observed in

579: the experiment does not stem from interference effects of the

580: phonons. In fact, the oscillations generated by this interference

581: occur on a much larger energy scale than the structure found in the

582: experimental curve. In contrast our results show that the bump on the

583: emission side of the current spectrum is due to another

584: mechanism: on the one hand we see in Fig.~\ref{ist_exprtrgfig}

585: that for small $\epsilon>0$ the inelastic

586: current grows with increasing $\epsilon$. On the other hand however,

587: the electron densities are simultaneously sharpened, so that the

588: cutoff $D$ is increased. For larger $\epsilon$ this again reduces the

589: inelastic current.

590: 

591: In summary, we have applied the RTRG method to the coupled quantum dot

592: system in a phonon bath in nonequilibrium. By accounting for 

593: both the coupling to the leads and the coupling to the environmental

594: phonons nonperturbatively we achieved a

595: reliable solution for

596: the stationary tunnel current. For the first time both the elastic and

597: the inelastic current of the experiment could

598: quantitatively be reproduced. Our analysis shows the importance of

599: the finite width $\sigma$ of the electron densities within one dot,

600: and for the experiment, the dependence of $\sigma$ on the energy

601: difference $\epsilon$

602: between the dots was calculated.

603: 

604: {\it Acknowledgments}.

605: We acknowledge useful discussions with T. Brandes and

606: K. Sch\"onhammer. We also thank

607: T. Fujisawa for the experimental data. This work was supported by the

608: ``Deutsche

609: Forschungsgemeinschaft'' as part of ``SFB 345'' (M.K.) and ``SFB 195''

610: (H.S.).

611: 

612: \vspace{1cm}

613: \begin{appendix}

614: \begin{center}{\bf Appendix}\end{center}

615: 

616: To obtain the explicit form of the RG equations one has to make a

617: choice of the $t_c$ dependence of the bath contractions

618: $\gamma^{pp'}_{\mu\mu'}(t,t_c)$.

619: Since the bosonic bath contractions $\gamma^{pp'}_{b_jb_k}(t)$

620: ($j,k=1,2,3$) correspond to those of the spin-boson model, we choose the

621: $t_c$ dependence as in Refs.~\cite{mk-hs_sbm,mk-hs_sbmproc}:

622: \begin{eqnarray}

623: \gamma^{pp'}_{b_jb_k}(t,t_c) &=& \frac{d}{dt}\left({\tilde R}_{jk}(t)\Theta(t-t_c)\right)\nonumber\\

624: && + ip'S_{jk}(t)\Theta(t-t_c)\,.

625: \label{gammabtceq}

626: \end{eqnarray}

627: The functions ${\tilde R}_{jk}(t)$ and $S_{jk}(t)$ are defined by

628: \begin{equation}

629: \label{gammarseq}

630: \gamma^{pp'}_{b_jb_k}(t) = \frac{d}{dt}{\tilde R}_{jk}(t) + ip'S_{jk}(t)\,.

631: \end{equation}

632: From Eqs.~(\ref{alphaq1eq}) - (\ref{gammaq1eq}) we obtain

633: \begin{eqnarray}

634: {\tilde R}_{11}(t) &=& g \,{\rm Re}\Bigl[\pi T\coth\left(\pi

635:  T(t-i/D)\right)\Bigr.\nonumber\\

636: &&\Bigl. +

637: \frac{\omega_d}{2}\ln\left(\frac{\sinh\left(\pi

638: T(t-1/\omega_d-i/D)\right)}{\sinh\left(\pi

639: T(t+1/\omega_d-i/D)\right)}\right)\Bigr]\label{r11eq}\,,\\

640: \label{s11eq}

641: S_{11}(t) &=& g \,{\rm

642: Im}\Bigl[\frac{1/\omega_d^2}{\left((t-i/D)^2-1/\omega_d^2\right)\left(t-i/D\right)^2}\Bigr]\,,\\

643: \label{rs12eq}

644: {\tilde R}_{12} &=& {\tilde R}_{21} = S_{12} = S_{21} = 0\,,\\

645: \label{rs13eq}

646: {\tilde R}_{13} &=& {\tilde R}_{31} = S_{13} = S_{31} = 0\,,\\

647: \label{r22eq}

648: {\tilde R}_{22}(t) &=& \frac{g}{2}e^{-D/\omega_d} \,{\rm Re}\Bigl[\pi T\coth\left(\pi T(t-i/D)\right)\Bigr]\,,\\

649: \label{s22eq}

650: S_{22}(t) &=& -\frac{g}{2}e^{-D/\omega_d} \,{\rm

651: Im}\Bigl[\frac{1}{(t-i/D)^2}\Bigr]\,,\\

652: {\tilde R}_{23}(t) &=& {\tilde R}_{32}(t) = -g \,{\rm

653: Re}\Bigl[\omega_d e^{-D/2\omega_d} \Bigr.\nonumber\\

654: && \Bigl.\times\ln\left(\frac{\sinh\left(\pi

655: T(t-1/2\omega_d-i/D)\right)}{\sinh\left(\pi

656: T(t+1/2\omega_d-i/D)\right)}\right)\Bigr]\label{r23eq}\,,\\

657: \label{s23eq}

658: S_{23}(t) &=& S_{32}(t) = -g \,{\rm

659: Im}\Bigl[\frac{e^{-D/2\omega_d}}{\left((t-i/D)^2-1/4\omega_d^2\right)}\Bigr]\,,\\{\tilde R}_{33}(t) &=& g \,{\rm Re}\Bigl[\pi T\coth\left(\pi

660:  T(t-i/D)\right)\Bigr.\nonumber\\

661: &&\Bigl. -

662: \frac{\omega_d}{2}\ln\left(\frac{\sinh\left(\pi

663: T(t-1/\omega_d-i/D)\right)}{\sinh\left(\pi

664: T(t+1/\omega_d-i/D)\right)}\right)\Bigr]\label{r33eq}\,,\\

665: \label{s33eq}

666: S_{33}(t) &=& g \,{\rm

667: Im}\Bigl[\frac{1/\omega_d^2 - 2(t-i/D)^2}{\left((t-i/D)^2-1/\omega_d^2\right)\left(t-i/D\right)^2}\Bigr]\,.

668: \end{eqnarray}

669: Here, we assumed that the high-energy cutoff $D$ is much larger than

670: the temperature $T$. In Eqs.~(\ref{r22eq}) - (\ref{s23eq}) one again

671: recognizes that the influence of the

672: offdiagonal interaction strongly depends on the width of the electron

673: densities $\sigma=|\vec d|\omega_d/D$.

674: The fermionic contractions can be written as

675: \begin{eqnarray}

676: \label{gammaffeq}

677: \gamma^{pp'}_{\eta f\eta'f'}(t) &=& \delta_{\eta,-\eta'}\delta_{f,f'}

678: \left(\frac{1}{2}\left(\gamma_{\eta f}(t)+\gamma_{-\eta

679: f}(-t)\right)\right.\nonumber\\

680: &&\left. +

681: \frac{p'}{2}\left(\gamma_{\eta f}(t)-\gamma_{-\eta f}(-t)\right)\right)\,,

682: \end{eqnarray}

683: where the indices $\eta$ and $f$ run over $\pm$ and $l,r$. 

684: We introduce

685: \begin{eqnarray}

686: \label{gammaleq}

687: \Gamma_l(\epsilon) &=& 2\pi \sum_k |V_k|^2\delta(\epsilon-\epsilon_k)\,,\\

688: \label{gammareq}

689: \Gamma_r(\epsilon) &=& 2\pi \sum_k |W_k|^2\delta(\epsilon-\epsilon_k)\,,

690: \end{eqnarray}

691: so that we obtain

692: \begin{equation}

693: \label{gammaetafeq}

694: \gamma_{\eta f}(t) =

695: \frac{-iT\Gamma_fe^{-i\eta\mu_ft}}{2\sinh\left(\pi

696: T(t-i/D_f)\right)}\,.

697: \end{equation}

698: Here, we introduced a bandwidth $D_f$ of the reservoir $f$ and assumed,

699: that $\Gamma_f(\epsilon)\approx{\rm const.}$ holds.

700: The cutoff-dependence of the fermionic contractions is chosen as

701: \begin{equation}

702: \label{gammaftceq} 

703: \gamma^{pp'}_{\eta f\eta'f'}(t,t_c) = \gamma^{pp'}_{\eta

704: f\eta'f'}(t)\Theta(t-t_c)\,.

705: \end{equation}

706: 

707: The final RG equations then read (for a more detailed derivation we

708: refer to Ref.~\cite{hs})

709: \begin{eqnarray}

710: \label{dsigmaeq}

711: \frac{d\Sigma_{(I)}}{dt_c} &=&

712:  \sum_{p_1,p_2,j,k}-i\left({\tilde R}_{jk}(t_c)A_{(I)b_j}^{p_1}(t_c)(L_0-z)\right.\nonumber\\

713: &&\left.\hspace{1cm} +

714: p_2S_{jk}(t_c)A_{(I)b_j}^{p_1}(t_c)\right)B_{b_k}^{p_2}\nonumber\\

715: &&\hspace{-0.5cm} - \sum_{p_1,p_2,\eta,f}\gamma^{p_1p_2}_{-\eta f\eta

716: f}(t_c)\hat\sigma^{p_1p_2}A_{(I)-\eta f}^{p_1}(t_c)B_{\eta f}^{p_2}\,,\\

717: \label{dleq}

718: \frac{dL_0}{dt_c} &=& \sum_{p_1,p_2,j,k}\left({\tilde R}_{jk}(t_c)[G_{b_j}^{p_1}(t_c),L_0]\right.\nonumber\\

719: &&\left.\hspace{1cm} +

720: p_2S_{jk}(t_c)G_{b_j}^{p_1}(t_c)\right)G_{b_k}^{p_2}\nonumber\\

721: &&\hspace{-0.5cm} - i\sum_{p_1,p_2,\eta,f}\gamma^{p_1p_2}_{-\eta f\eta

722: f}(t_c)\hat\sigma^{p_1p_2}G_{-\eta f}^{p_1}(t_c)G_{\eta f}^{p_2}\,,\\

723: \label{dgeq}

724: \frac{dG_{\mu}^p}{dt_c} &=& \sum_{p_1,p_2,j,k}

725: {\tilde R}_{jk}(t_c)\left(G_{b_j}^{p_1}(t_c)G_{\mu}^p-G_{\mu}^pG_{b_j}^{p_1}(t_c)\right)G_{b_k}^{p_2}\nonumber\\

726: && -

727: i\int_0^{t_c}dt\,\left(G_{b_j}^{p_1}(t)G_{\mu}^p-G_{\mu}^pG_{b_j}^{p_1}(t)\right)\nonumber\\

728: && \times \left({\tilde R}_{jk}(t_c)[L_0,G_{b_k}^{p_2}(t-t_c)]\right.\nonumber\\

729: &&\left.\hspace{1cm} +

730: p_2S_{jk}(t_c)G_{b_k}^{p_2}(t-t_c)\right)\nonumber\\

731: && + \sum_{p_1,p_2,\eta,f}\gamma^{p_1p_2}_{-\eta f\eta

732: f}(t_c)\int_0^{t_c}dt\, \left(G_{\mu}^p\hat\sigma^{p_1p_2}G_{-\eta

733: f}^{p_1}(t)\right.\nonumber\\

734: &&\left.\hspace{0.0cm} - \eta^{pp_2}_{\mu}\hat\sigma^{p_1p_2}G_{-\eta

735: f}^{p_1}(t)G_{\mu}^p\right)G_{\eta f}^{p_2}(t-t_c)\,,\\

736: \label{daeq}

737: \frac{dA_{(I)\mu}^p}{dt_c} &=& \sum_{p_1,p_2,j,k}

738: {\tilde R}_{jk}(t_c)\left(A_{(I)b_j}^{p_1}(t_c)G_{\mu}^p\right.\nonumber\\

739: &&\left.\hspace{1cm} - A_{(I)\mu}^pG_{b_j}^{p_1}(t_c)\right)G_{b_k}^{p_2}\nonumber\\

740: && -

741: i\int_0^{t_c}dt\,\left(A_{(I)b_j}^{p_1}(t)G_{\mu}^p-A_{(I)\mu}^pG_{b_j}^{p_1}(t)\right)\nonumber\\

742: && \times \left({\tilde R}_{jk}(t_c)[L_0,G_{b_k}^{p_2}(t-t_c)]\right.\nonumber\\

743: &&\left.\hspace{1cm} + p_2S_{jk}(t_c)G_{b_k}^{p_2}(t-t_c)\right)\nonumber\\

744: && + \sum_{p_1,p_2,\eta,f}\gamma^{p_1p_2}_{-\eta f\eta

745: f}(t_c)\nonumber\\

746: && \times\int_0^{t_c}dt\, \left(A_{(I)\mu}^p\hat\sigma^{p_1p_2}G_{-\eta

747: f}^{p_1}(t)\right.\nonumber\\

748: &&\left.\hspace{0.0cm} - \eta^{pp_2}_{\mu}\hat\sigma^{p_1p_2}A_{(I)-\eta

749: f}^{p_1}(t)G_{\mu}^p\right)G_{\eta f}^{p_2}(t-t_c)\,,\\

750: \label{dbeq}

751: \frac{dB_{\mu}^p}{dt_c} &=& \sum_{p_1,p_2,j,k}

752: {\tilde R}_{jk}(t_c)G_{b_j}^{p_1}(t_c)G_{\mu}^pB_{b_k}^{p_2} -

753: i\int_0^{t_c}dt\,G_{b_j}^{p_1}(t)\nonumber\\

754: && \times  G_{\mu}^p\left({\tilde R}_{jk}(t_c)(L_0-z)B_{b_k}^{p_2}(t-t_c)\right.\nonumber\\

755: &&\left.\hspace{1cm} + p_2S_{jk}(t_c)B_{b_k}^{p_2}(t-t_c)\right)\nonumber\\

756: && - \sum_{p_1,p_2,\eta,f}\gamma^{p_1p_2}_{-\eta f\eta

757: f}(t_c)\nonumber\\

758: && \times\int_0^{t_c}dt\, \eta^{pp_2}_{\mu}\hat\sigma^{p_1p_2}G_{-\eta

759: f}^{p_1}(t)G_{\mu}^pB_{\eta f}^{p_2}(t-t_c)\,.

760: \end{eqnarray}

761: The interaction picture is defined by $G_{\mu}^p(t)=e^{iL_0t}G_{\mu}^p

762: e^{-iL_0t}$,

763: $A_{(I)\mu}^p(t)=e^{izt}A_{(I)\mu}^p e^{-iL_0t}$, and

764: $B_{\mu}^p(t)=e^{iL_0t}B_{\mu}^p e^{-izt}$. 

765: The function $\eta^{pp'}_{\mu}$ and the superoperator

766: $\hat\sigma^{pp'}$ account for additional signs arising from the

767: commutation of fermionic field operators.

768: They are given by

769: \begin{eqnarray}

770: \label{etaeq}

771: \eta^{pp'}_{\mu} &=& \left\{\begin{array}{cc} -pp' & \mbox{for $\mu$

772: fermionic}\\

773: 1 & \mbox{else}

774: \end{array}\right.\,,\\

775: \label{sigmaeq}

776: \left(\hat\sigma^{pp'}\right)_{ss',ss'} &=& \left\{\begin{array}{cc}

777: pp' & \mbox{for $N_s-N_{s'}={\rm odd}$}\\

778: 1 & \mbox{for $N_s-N_{s'}={\rm even}$}

779: \end{array}\right.\,.

780: \end{eqnarray}

781: We also note, that in the limit of large $D_l$ and $D_r$ the first

782: part of the fermionic contractions in Eq.~(\ref{gammaffeq}) can be

783: written as

784: \begin{equation}

785: \label{gammadeltaeq}

786: \frac{1}{2}\left(\gamma_{\eta f}(t)+\gamma_{-\eta

787: f}(-t)\right) = \frac{\Gamma_f}{2}\delta(t)\,.

788: \end{equation}

789: Thus, for $D_l,D_r\rightarrow\infty$, this contribution to the

790: differential equations

791: Eqs.~(\ref{dsigmaeq}) - (\ref{dbeq}) can be

792: incorporated in the initial conditions.

793: \end{appendix}

794: 

795: \begin{references}

796: 

797: \bibitem{raikh}

798: M. Raikh, A. Asenov, Superlatt. Microstr. {\bf 11}, 325 (1992).

799: 

800: \bibitem{tarucha}

801: S. Tarucha, D. G. Austing, Y. Tokura, W. G. van der Wiel,

802: and L. P. Kouwenhoven, Phys. Rev. Lett. {\bf 84}, 2485

803: (2000).

804: 

805: \bibitem{cronenwett}

806: S. M. Cronenwett, T. H. Oosterkamp, L. P. Kouwenhoven, Science {\bf

807: 281}, 540 (1998).

808: 

809: \bibitem{wiel}

810: W.G. van der Wiel, S. De Franceschi, T. Fujisawa, J. M. Elzerman,

811: S. Tarucha, L.P. Kouwenhoven, Science {\bf 289}, 2105 (2000).

812: 

813: \bibitem{sasaki}

814: S. Sasaki, S. De Franceschi, J. M. Elzerman, W. G. van der Wiel,

815: M. Eto, S. Tarucha, L. P. Kouwenhoven, Nature {\bf 405}, 764-767 (2000).

816: 

817: \bibitem{loss}

818: D. Loss and D. P. DiVincenzo, Phys. Rev. A {\bf 57}, 120

819: (1998).

820: 

821: \bibitem{fujisawa}

822: T. Fujisawa, T. H. Oosterkamp, W. G. van der Wiel, B. W. Broer,

823: R. Aguado, S. Tarucha, L. P. Kouwenhoven, Science {\bf 282}, 932

824: (1998).

825: 

826: \bibitem{brandes}

827: T. Brandes, B. Kramer, Phys. Rev. Lett. {\bf 83}, 3021 (1999).

828: 

829: \bibitem{leggett-etal}

830: A. J. Leggett, S. Chakravarty, A. T. Dorsey, M. P. A. Fisher, A. Garg,

831: W. Zwerger, Rev. Mod. Phys. {\bf 59}, 1 (1987).

832: 

833: \bibitem{weiss-book}

834: U. Weiss, {\it Quantum Dissipative Systems} (World

835: Scientific, Singapore, 2000).

836: 

837: \bibitem{koenig-hs}

838: J. K\"onig and H. Schoeller, Phys. Rev. Lett. {\bf 81}, 3511

839: (1998).

840: 

841: \bibitem{mk-hs_pol}

842: M. Keil, H. Schoeller, Phys. Rev. B {\bf 62}, 2990 (2000).

843: 

844: \bibitem{hs-koenig}

845: H.Schoeller and J.K\"onig,  

846: Phys. Rev. Lett. {\bf 84}, 3686 (2000).

847: 

848: \bibitem{mk-hs_sbm}

849: M. Keil, H. Schoeller, Phys. Rev. B {\bf 63}, 180302(R) (2001).

850: 

851: \bibitem{mk-hs_sbmproc}

852: M. Keil, H. Schoeller, Chemical Physics {\bf 268}, 11-20 (2001).

853: 

854: \bibitem{hs}

855: H.Schoeller,

856: in {\it Low-Dimensional Systems}, ed. T.Brandes

857: (Springer, 1999), p.137.

858: 

859: \bibitem{bruus}

860: H. Bruus, K. Flensberg, and H. Smith, Phys. Rev. B {\bf 48}, 11144 (1993).

861: 

862: \bibitem{stoof}

863: T. H. Stoof and Y. V. Nazarov, Phys. Rev. B {\bf 53}, 1050 (1996).

864: 

865: \end{references}

866: 

867: \begin{figure}

868: \centerline{\psfig{figure=dd.eps,width=8cm,height=6cm,angle=0}}

869: \vspace{0.5cm}

870: \caption{The double quantum dot may be charged with one additional

871: electron in the left or right dot. The corresponding states

872: $|l\rangle$ and $|r\rangle$ are coupled by the tunnel amplitude

873: $T_c$. The couplings to the leads are given by $\Gamma_{l(r)}$. The

874: energy difference between the quantum dots is

875: $\epsilon=\epsilon_l-\epsilon_r$, and there is an external voltage

876: $V=\mu_l-\mu_r$. The interaction with the acoustical phonons (dashed

877: lines), consists of a diagonal part with the coefficients $\alpha_q$

878: and $\beta_q$, and an

879: offdiagonal part with the constants $\gamma_q$.}

880: \label{ddfig}

881: \end{figure}

882: 

883: \newpage

884: 

885: \begin{figure}

886: \centerline{\psfig{figure=it.eps,width=8cm,height=1.8cm,angle=0}}

887: \vspace{0.5cm}

888: \caption{Diagrammatic expression for $\langle I\rangle (t)$. The two

889: lines of the Keldysh contour are put together to one line. The

890: irreducible diagrams in $\Sigma_I$ include the leftmost vertex

891: superoperator at the time point $t'$ and the current superoperator

892: $A_I$ at the time point $t$. $\Sigma_I$ acts on $p(t')$, which is

893: represented by the horizontal line.}

894: \label{itfig}

895: \end{figure}

896: 

897: \begin{figure}

898: \centerline{\psfig{figure=ist_brartrg.eps,width=8cm,height=5cm,angle=0}}

899: \caption{Stationary tunnel current as a function of $\epsilon$ for

900: $T_c=\Gamma_l=\Gamma_r=1\mu eV$ and $D_l=D_r=1meV$. and $D=100\mu eV$

901: respectively $D=150\mu eV$ . Solid

902: line: $D=150\mu eV$. Dashed line: $D=150\mu eV$, $\gamma_q=0$. Dotted

903: line: $D=100\mu eV$. Dot-dashed line: $D=100\mu eV$, $\gamma_q=0$.}

904: \label{ist_brartrgfig}

905: \end{figure}

906: 

907: \begin{figure}

908: \centerline{\psfig{figure=ist_expstoof.eps,width=8cm,height=5cm,angle=0}}

909: \caption{Stationary tunnel current as a function of $\epsilon$. Solid

910: line: Experiment. Dashed line: Stoof-Nazarov result

911: for the case of no electron-phonon interaction, with the parameters

912: $T_c=0.124\mu eV$ and $\Gamma_l=\Gamma_r=3.5\mu eV$.}

913: \label{ist_expstooffig}

914: \end{figure}

915: 

916: \begin{figure}

917: \centerline{\psfig{figure=ist_exprtrg.eps,width=7cm,height=5cm,angle=0}}

918: \caption{Stationary tunnel current as

919: a function of $\epsilon$. Solid

920: line: Experiment. Dashed line: RTRG with $T_c=0.375\mu

921: eV$, $\Gamma_l=\Gamma_r=3.5\mu

922: eV$, $D_l=D_r=1meV$ and $D=70\mu eV$. Dotted line: RTRG with $T_c=0.375\mu

923: eV$, $\Gamma_l=\Gamma_r=3.5\mu

924: eV$, $D_l=D_r=1meV$ and $D=100\mu eV$.}

925: \label{ist_exprtrgfig}

926: \end{figure}

927: 

928: \begin{figure}

929: \centerline{\psfig{figure=sig_rtrg.eps,width=8cm,height=5cm,angle=0}}

930: \caption{The width of the electron density within one dot, $\sigma$, as

931: a function of $\epsilon$. $T_c=0.375\mu eV$, $\Gamma_l=\Gamma_r=3.5\mu

932: eV$ and $D_l=D_r=1meV$.}

933: \label{sig_rtrgfig}

934: \end{figure}

935: 

936: \end{document}

937: