1: % Template article for preprint document class `elsart'

2: % SP 2001/01/05

3: 

4: % Use the option doublespacing or reviewcopy to obtain double line spacing

5: %\documentclass[doublespacing]{elsart}

6: \documentclass{elsart}

7: 

8: % The amssymb package provides various useful mathematical symbols

9: %\usepackage{amssymb}

10: \usepackage{amsmath}

11: \usepackage{graphicx}

12: 

13: \graphicspath{{img/}}

14: 

15: \makeatletter

16: \def\@date{22 May 2002}

17: \makeatother

18: 

19: \renewcommand{\l}{\ell}

20: \renewcommand{\L}{L}%\mathcal{L}}

21: 

22: \begin{document}

23: 

24: \begin{frontmatter}

25: \title{Laser controlled molecular switches and transistors}

26: \author{J\"org Lehmann},

27: \author{S\'ebastien Camalet},

28: \author{Sigmund Kohler}, and

29: \author{Peter H\"anggi\corauthref{cor1}}

30: \corauth[cor1]{Corresponding author. Tel.: +49-821-598-3250; 

31: fax: +49-821-598-3222; e-mail: Peter.Hanggi@physik.uni-augsburg.de}

32: \address{Institut f\"ur Physik, Universit\"at Augsburg, 

33:          Universit\"atsstra{\ss}e~1, D-86135 Augsburg, Germany}

34: 

35: \begin{abstract}

36: We investigate the possibility of optical current control through single molecules

37: which are weakly coupled to leads.  A master equation approach for the transport

38: through a molecule is combined with a Floquet theory for the time-dependent

39: molecule.  This yields an efficient numerical approach to the evaluation of

40: the current through time-dependent nano-structures in the presence of a finite

41: external voltage.  We propose tunable optical current switching in two- and

42: three-terminal molecular electronic devices driven by properly adjusted laser

43: fields, \textit{i.e.} a novel class of molecular transistors.

44: 

45: \end{abstract}

46: 

47: \begin{keyword}

48:   molecular electronics \sep quantum control 

49:   \PACS 85.65.+h \sep 33.80.-b \sep 73.63.-b \sep 05.60.Gg

50: \end{keyword}

51: \end{frontmatter}

52: 

53: \section{Introduction}

54: 

55: Spurred by the ongoing experimental progress in the field of molecular

56: electronics \cite{Reed1997a,Cui2001a,Joachim2000a,Schon2001a,Reichert2002a},

57: the theoretical interest in transport properties of molecules has revived

58: \cite{Nitzan2001a}.  Tight-binding models for the wire have been used to

59: compute current-voltage characteristics, within a scattering approach

60: \cite{Mujica1994a} and from electron transfer theory \cite{Nitzan2001a}.  Both

61: approaches bear the same essentials \cite{Nitzan2001b}.  For high temperatures,

62: the wire electrons loose their quantum coherence and the transport is dominated

63: by incoherent hopping between neighbouring sites \cite{Petrov2001a}.  Recently,

64: the current-voltage characteristics has been obtained from a quantum-chemical

65: \textit{ab initio} description of the molecule \cite{Heurich2001a}.  The

66: results were in good agreement with recent experiments \cite{Reichert2002a}.

67: 

68: Typical electronic excitation energies in molecules are in the range up to an

69: eV and, thus, correspond to light quanta from the optical and the infrared

70: spectral regime where most of today's lasers work.  It is therefore natural to

71: use such coherent light sources to excite molecules and to study their

72: influence on the transport properties aiming to find a way of manipulating

73: currents.  One particularly prominent example of quantum control is the

74: so-called coherent destruction of tunnelling (CDT), \textit{i.e.} the

75: suppression of the tunnelling dynamics in an \textit{isolated} bistable

76: potential by the purely coherent influence of an oscillating bias

77: \cite{Grossmann1991a,Grossmann1992a,Morillo1993a,Goychuk1996a,Grifoni1998a}.

78: The crucial point there is that the long-time dynamics in a periodically driven

79: quantum system is no longer dominated by the energies, but rather by the

80: so-called quasienergies \cite{Grifoni1998a,Shirley1965a,Sambe1973a}.  The

81: latter may be degenerate for properly chosen driving parameters yielding a

82: divergent time-scale.  Inspired by these results, we address in this Letter the

83: question of controlling by use of properly tailored laser fields the transport

84: through time-dependent \textit{open} systems, \textit{i.e.} systems that allow

85: for a particle exchange with external leads.

86: 

87: For the computation of electrical currents through wires exposed to strong

88: laser fields, we put forward Floquet approach \cite{Lehmann2002b}.  The central

89: idea of this method lies in a non-perturbative solution of the Schr\"odinger

90: equation of the isolated time-dependent wire plus laser field, while the

91: wire-lead coupling is treated perturbatively.  The resulting density operator

92: equation is decomposed into a time-dependent Floquet basis permitting a

93: numerically efficient treatment.  We generalise here this method to the

94: analysis of networks with an arbitrary number of contacts to leads.

95: Subsequently we apply the so obtained formalism to the investigation of optical

96: current switching in two- and three-terminal devices as prototypical examples

97: of a new class of molecular transistors.

98: 

99: \section{Model for wire and leads}

100: 

101: %%%% Figure 1 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

102: \begin{figure}

103:   \begin{center}

104:     \includegraphics[width=6truecm]{fig1.eps}

105:   \end{center}

106:   \caption{Molecular circuit consisting of $N=6$ sites of which the

107:     sites $1,\ldots,L$ are coupled to $\L=4$ leads.}

108:   \label{fig:system}

109: \end{figure}

110: %%%% Figure 1 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

111: 

112: We embark by specifying the model Hamiltonian of the entire system as

113: sketched in Fig.~\ref{fig:system}.  It consists of the molecule in the

114: laser field, ideal leads, and the molecule-leads coupling Hamiltonian,

115: \begin{equation}

116: H(t)=H_{\text{molecule}}(t) + H_{\text{leads}} +

117: H_{\text{molecule-leads}}\ .

118: \end{equation}

119: The irradiated molecule is modelled by a tight-binding description taking into

120: account $N$ molecular orbitals $|n\rangle$, which are relevant for the

121: transport.  Disregarding the electron-electron interaction, the most general

122: form of the Hamiltonian reads

123: \begin{equation}

124: \label{eq:Hmolecule} H_{\text{molecule}}(t)=\sum_{n,n'} H_{nn'}(t)\,

125: c_n^\dagger c_{n'},

126: \end{equation}

127: where the fermionic operators $c_n$ and $c_n^\dagger$ destroy and create,

128: respectively, an electron in the molecular orbital $|n\rangle$. The sums

129: extend over all tight-binding orbitals.  The $\mathcal{T}$-periodic

130: time-dependence of the single-particle Hamiltonian

131: $H_{nn'}(t)=H_{nn'}(t+\mathcal{T})$, reflects the influence of the laser field

132: with frequency $\Omega=2\pi/\mathcal{T}$.

133: %

134: The $L$ ideal leads are described by grand-canonical ensembles of electrons

135: at temperature $T$ and electro-chemical potential $\mu_\l$, $\l=1,\dots,L$.

136: Thus, the lead Hamiltonian reads

137: $H_\mathrm{leads} = \sum_{q\l} \epsilon_{q\l} \, c_{q\l}^\dagger c_{q\l}$,

138: where $c_{q\l}$ destroys an electron in state $q$ in lead $\l$.

139: All expectation values of lead operators can be traced back to

140: $\langle c_{q\l}^\dagger c_{q'\l'} \rangle = \delta_{qq'}

141: \delta_{\l\l'} f(\epsilon_{q\l}-\mu_\l)$, where $f(\epsilon)

142: =(1+\e^{\epsilon/k_B T})^{-1}$ denotes the Fermi function.

143: The model is completed by the molecule-leads tunnelling Hamiltonian

144: \begin{equation}

145: H_{\text{molecule-leads}}

146: =\sum_{q\l} V_{q\l} \, c_{q\l}^\dagger \, c_{\l} + \mathrm{h.c.}\ ,

147: \end{equation}

148: that connects each lead directly to one of the suitably labelled molecular

149: orbitals.  Since we are not interested here in the effects that arise from the

150: microscopic details of the molecule-lead coupling, we restrict our analysis in

151: the following to energy-independent couplings, \textit{i.e.} $\Gamma_{\l} = (2\pi/\hbar)

152: \sum_q |V_{q\l}|^2 \, \delta(\epsilon - \epsilon_{q\l})$.

153: 

154: 

155: \section{Perturbative description and Floquet ansatz}

156: 

157: Let us assume that the dynamics of the driven wire is dominated by the

158: time-dependent wire Hamiltonian so that the coupling to the leads can be taken

159: into account as a perturbation.  This allows to derive by use of standard

160: methods the approximate equation of motion for the total density operator

161: $\varrho(t)$, 

162: \begin{align}

163: \label{eq:master}

164: \dot\varrho(t) =

165: & -\frac{i}{\hbar}[H_{\rm molecule}(t)+H_{\rm leads},\varrho(t)] \\

166: & -\frac{1}{\hbar^2}\int\limits_0^\infty \!\mathrm{d}\tau

167:   [H_\mathrm{molecule-leads},[\widetilde{H}_\mathrm{molecule-leads}(t-\tau,t),

168:   \varrho(t)]] . \nonumber

169: \end{align}

170: We have omitted a transient term that depends purely on the initial preparation.

171: The tilde denotes operators in the interaction picture with respect to the

172: molecule plus the lead Hamiltonian, $\widetilde{O}(t,t')=U_0^\dagger(t,t') O(t)

173: U_0(t,t')$, where $U_0(t,t')$ is the time-evolution operator without the

174: coupling.  The first term describes the coherent dynamics of the electrons on

175: the wire, while the second term represents incoherent hopping of electrons

176: between the leads and the wire.

177: 

178: The net (incoming minus outgoing) electrical current that flows from lead $\l$

179: into the molecule is then given by the rate at which the electron number in the

180: corresponding lead decreases multiplied by the electron charge $-e$,

181: \begin{equation}

182:   \label{eq:current}

183:   I_\l(t) = e\, \frac{\mathrm{d}}{\mathrm{d}t}\langle N_\l\rangle\ .

184: \end{equation}

185: Note that this expectation value is time-dependent through the non-equilibrium

186: density operator $\varrho(t)$.  To evaluate the right-hand side of

187: Eq.~(\ref{eq:current}), we employ Eq.~\eqref{eq:master} to derive after some

188: algebra the result

189: \begin{equation}

190: \label{eq:current_general}

191: \begin{split}

192: I_\l(t)  = 

193:  -e   \Gamma_{\l} \bigg\{ &\mathop{\rm Re} 

194: \int\limits_0^\infty \frac{\mathrm{d}\tau}{\hbar} \! 

195: \int  \!\frac{\mathrm{d}\epsilon}{\pi}\,

196: e^{i\epsilon\tau/\hbar}          

197: f(\epsilon-\mu_\l)\big\langle[c_\l,

198: \tilde c_{\l}^\dagger(t-\tau,t)]_+\big\rangle \\

199:  &- \big\langle c_\l^\dagger

200:  c_{\l}\big\rangle\bigg\}\ .

201: \end{split}

202: \end{equation}

203: This expression still contains the yet unknown expectation values of solely

204: those wire operators with a direct connection to lead $\l$.  It depends in

205: particular on the Heisenberg operators $\tilde c_\l^\dagger(t-\tau,t)$ and thus

206: implicitly on the dynamics of the driven wire.  Let us therefore focus on the

207: single-particle dynamics of the periodically time-dependent wire Hamiltonian

208: $H_{nn'}(t)$.  An established procedure for the solution of the corresponding

209: Schr\"odinger equation is to employ a Floquet ansatz amounting to a

210: non-perturbative treatment of the external driving.  There one uses the fact

211: that a complete set of solutions is of the form

212: $|\Psi_\alpha(t)\rangle=\exp(-i\epsilon_\alpha t/\hbar)|\Phi_\alpha(t) \rangle$.

213: The so-called quasienergies $\epsilon_\alpha$ take over the role of the

214: energy eigenvalues in static systems and govern the long-time dynamics.  The Floquet

215: modes $|\Phi_\alpha(t)\rangle$ obey the time-periodicity of the driving field

216: which allows to express them as a Fourier series, $|\Phi_\alpha(t)\rangle =

217: \sum_{k=-\infty}^{\infty} \exp(-ik\Omega t) |\Phi_{\alpha,k}\rangle$.  They can

218: be obtained from the eigenvalue equation

219: \cite{Shirley1965a,Sambe1973a,Grifoni1998a}

220: \begin{equation}

221: \label{floquet_hamiltonian}

222: \Big(\sum_{n,n'}|n\rangle H_{nn'}(t) \langle n'|

223: -i\hbar\frac{\mathrm{d}}{\mathrm{d}t}\Big)

224: |\Phi_{\alpha}(t)\rangle = \epsilon_\alpha |\Phi_{\alpha}(t)\rangle

225: \ .

226: \end{equation}

227: Moreover, the Floquet modes define the complete set of operators

228: \begin{equation}

229: \label{eq:c_alpha}  

230: c_\alpha(t) = \sum_n \langle\Phi_\alpha(t)|n\rangle\, c_n  ,

231: \end{equation}

232: whose time-evolution assumes the convenient form $\tilde{c}_\alpha(t-\tau,t) =

233: \exp(i\epsilon_\alpha\tau/\hbar) c_\alpha(t)$.  The orthogonality of the Floquet

234: states at equal times \cite{Shirley1965a,Sambe1973a,Grifoni1998a} yields the

235: back-transformation $c_n=\sum_\alpha\langle n|\Phi_\alpha(t)\rangle$ and thus

236: results in the required spectral decomposition.

237: Using (\ref{eq:c_alpha}) and performing the energy and the $\tau$-integration

238: in Eq.~(\ref{eq:current_general}), we obtain for the time-averaged current

239: the main result

240: \begin{equation}

241: \label{eq:meancurrent}

242: \begin{split}

243: \bar{I}_\l = 

244: -\frac{e\Gamma_\l}{\hbar}\sum_{\alpha k}

245: \Big[&

246: \langle \Phi_{\alpha, k}|\l\rangle \langle \l|\Phi_{\alpha, k}\rangle 

247:   f(\epsilon_\alpha+k\hbar\Omega-\mu_\l) 

248: \\ &

249: -\sum_{\beta k'}

250:  \langle \Phi_{\alpha, k'+k} |\l\rangle\langle \l|\Phi_{\beta k'} \rangle

251:  R_{\alpha\beta,k}

252: \Big] .

253: \end{split}

254: \end{equation}

255: Here, we have introduced the expectation values $R_{\alpha\beta}(t)= \langle

256: c^\dagger_\alpha(t)c_{\beta}(t)\rangle$, which assume in the long-time limit the

257: time-periodicity of the driven system and thus can be decomposed into the

258: Fourier series $R_{\alpha\beta}(t)= \sum_k \exp(-ik\Omega t) R_{\alpha\beta,k}$.

259: It is straightforward to derive for the $R_{\alpha\beta,k}$ from the

260: density operator equation~(\ref{eq:master}) the following set of inhomogeneous linear

261: equations

262: \begin{align}

263: \label{eq:mastereq_fourier}

264: \lefteqn{\frac{i}{\hbar}(\epsilon_\alpha-\epsilon_\beta+k \hbar \Omega)R_{\alpha\beta,k}}

265: \\ && =

266:  \frac{1}{2}\sum_{\l k'} \Gamma_\l

267:     \Big\{ &

268:     \sum_{\beta'k''}

269:                 \langle\Phi_{\beta,k''+k'}|\l\rangle

270:                 \langle \l|\Phi_{\beta',k''+k}\rangle

271:                 R_{\alpha\beta',k'}

272: \nonumber \\&& {}+{} &

273:     \sum_{\alpha'k''}

274:                 \langle\Phi_{\alpha',k''+k'}|\l\rangle

275:                 \langle \l|\Phi_{\alpha,k''+k}\rangle

276:                 R_{\alpha'\beta,k'}

277: \nonumber \\ && {}-{} &

278:     \langle\Phi_{\beta,k'-k}|\l\rangle

279:     \langle \l|\Phi_{\alpha,k'}\rangle

280:     f(\epsilon_\alpha+k'\Omega-\mu_\l)

281: \nonumber \\ && {}-{} &

282:     \langle\Phi_{\beta,k'}|\l\rangle

283:     \langle \l|\Phi_{\alpha,k'+k}\rangle

284:     f(\epsilon_\beta+k'\Omega-\mu_\l)

285:    \Big\}

286: \nonumber\ ,

287: \end{align}

288: which will be solved numerically.  We have found that even in the case of

289: strong driving where the Floquet states comprise many sidebands, a few Fourier

290: coefficients $R_{\alpha\beta,k}$ are in fact sufficient to obtain numerical

291: convergence.  This justifies \textit{a posteriori} the use

292: of the Floquet states as a basis set.

293: 

294: To conclude the technical part of this work, we note that our approach goes

295: beyond a linear response treatment of the driving and additionally does not use

296: a so-called rotating wave approximation (RWA) \cite{Blumel1991a,Bruder1994a},

297: which neglects the oscillatory contributions to $R_{\alpha\beta}(t)$ by the

298: ansatz $R_{\alpha\beta,k}=P_\alpha\,\delta_{\alpha,\beta}\,\delta_{k,0}$.  In

299: fact, we found that a RWA solution delivers inaccurate results in the vicinity

300: of quasienergy degeneracies.

301: 

302: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

303: \section{Optical current gate}

304: 

305: As a first setup that may be suitable as a current control device, we

306: investigate the transport through a two-level system, \textit{i.e.} a wire that consists

307: of $N=2$ sites --- one of them is coupled to the left lead and the other to

308: the right lead.  Then the time-dependent wire Hamiltonian reads in the basis of

309: the molecular orbitals

310: \begin{equation}

311:   \label{eq:twosites}

312:     H_\mathrm{molecule}(t) = 

313:     \begin{pmatrix}

314:       E_\mathrm{L} && -\Delta\\

315:       -\Delta      && E_\mathrm{R}

316:     \end{pmatrix}

317:     + A 

318:     \begin{pmatrix}

319:       1 && 0\\

320:       0 && -1

321:     \end{pmatrix}\cos(\Omega t)\ ,

322: \end{equation}

323: where $\Delta$ denotes the tunnel matrix element between the two sites and

324: $E_\mathrm{L}$ and $E_\mathrm{R}$ are the corresponding on-site energies.  The

325: laser field contributes to the Hamiltonian (\ref{eq:twosites}) a time-dependent

326: bias with amplitude $A=-e\mathcal{E}d$, \textit{i.e.} charge times electrical

327: field strength times the site-to-site distance.  Note that the electrical field

328: may be drastically enhanced due to the presence of the metallic tips

329: \cite{Demming1998a}.  The effective coupling to each lead is assumed to be

330: equal, $\Gamma_{\l} = \Gamma$, and an external voltage $V$ is taken into account

331: by a difference in the electro-chemical potentials,

332: $\mu_\mathrm{L}-\mu_\mathrm{R}=-eV$.

333: 

334: We use in all numerical calculations the tunnel matrix element $\Delta$ as the

335: energy unit and assume that the effective couplings to the leads are by one order of

336: magnitude smaller, $\hbar\Gamma=0.1\Delta$. This corresponds to

337: a large contact resistance and ensures the applicability of a perturbational

338: approach.  A realistic value is $\Delta=0.1\mathrm{eV}$, resulting in a current

339: unit $e\Gamma=0.256\,\mathrm{\mu A}$.  For a site-to-site distance of $10${\AA}

340: and a laser frequency $\Omega=10\Delta/\hbar$, the driving amplitudes

341: considered below correspond to an electrical field amplitude of

342: $10^6\,\mathrm{V}/\mathrm{cm}$ at $1\,\mu\mathrm{m}$ wavelength.

343: 

344: %%%% Figure 2 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

345: \begin{figure}

346:   \begin{center}

347:     \includegraphics[width=7.5truecm]{fig2.eps}

348:   \end{center}

349:   \caption{Average current and quasienergy spectrum versus driving amplitude

350:     for a wire which consists of two sites between two electrodes (cf.\ inset)

351:     for unbiased ($E_\mathrm{R}=E_\mathrm{L}=0$, solid lines) and

352:     biased ($E_\mathrm{R}=-E_\mathrm{L}=0.1\Delta$, dashed lines)

353:     on-site energies.

354:     The leads' chemical potentials are $\mu_\mathrm{R}=-\mu_\mathrm{L}=10 \Delta$;

355:     the other parameters read $\hbar\Omega=10\Delta$, $k_B T=0.25\Delta$, 

356:     $\hbar\Gamma=0.1\Delta$.

357:     }

358:   \label{fig:twolevel}

359: \end{figure}%

360: %%%% Figure 2 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

361: The time-averaged current $\bar{I}=\bar{I}_\mathrm{L}=-\bar{I}_\mathrm{R}$

362: through the molecule in a case where both on-site energies are equal is

363: depicted in Fig.~\ref{fig:twolevel}.  As a striking feature, we find that at

364: certain values of the driving amplitude, the current collapses to less than 1\%

365: of its maximal value reached in the absence of the driving.  Closer inspection

366: (not shown) reveals that the width of this depression is proportional to the

367: molecule-lead coupling $\Gamma$.  Comparison with the quasienergy spectrum in

368: the lower panel demonstrates that the current break downs occur at quasienergy

369: crossings.  This relates the present phenomenon to the CDT, \textit{i.e.} the

370: standstill of the tunnel dynamics in a driven bistable potential at quasienergy

371: crossings \cite{Grossmann1991a}.  For the \textit{isolated} two-level system

372: (\ref{eq:twosites}) with $\Delta\ll\Omega,A$, the CDT condition reads

373: $\mathrm{J}_0(2A/\hbar\Omega)=0$ \cite{Grossmann1992a}, \textit{i.e.} the suppression

374: of the tunnelling dynamics is related to the zeros of the Bessel function

375: $\mathrm{J}_0$.  As our analysis reveals, the same condition results in a

376: suppression of the transport through the \textit{open} system.

377: 

378: An external voltage may be of peculiar influence to the on-site energies of a

379: molecular wire \cite{Mujica2000a} and cause an effective bias $E_\mathrm{L}\neq

380: E_\mathrm{R}$ in originally symmetric molecules.  Thus, a crucial question is

381: whether the above current suppression is stable against such a modification.

382: The broken lines in Fig.~\ref{fig:twolevel} demonstrate that this is indeed the

383: case.  Although the quasienergies now form an avoided crossing, the current

384: breakdowns do survive; they are even more pronounced, but slightly shifted

385: towards larger driving amplitudes.  This robustness of CDT based current

386: control combined with the huge on/off ratio suggests the presented setup as a

387: promising alternative to structural chemistry-based switching devices

388: \cite{Chen1999a,Collier2000a}.

389: 

390: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

391: \section{Optical current router}

392: 

393: An experimentally more ambitious configuration consists in a planar molecule

394: with $N=4$ sites, three of which are coupled to a central site and are

395: directly connected to leads (cf.\ inset of Fig.~\ref{fig:switch1}).  We borrow

396: from electrical engineering the designation $\mathrm{E}$, $\mathrm{C}_1$, and

397: $\mathrm{C}_2$.  Here, an external voltage is always applied such that

398: $\mathrm{C}_1$ and $\mathrm{C}_2$ have equal electro-chemical potential,

399: \textit{i.e.} $\mu_{\mathrm{C}_1}=\mu_{\mathrm{C}_2}\neq\mu_\mathrm{E}$.  In a perfectly

400: symmetric molecule, where all on-site energies equal each other, reflection

401: symmetry at the horizontal axis ensures that any current which enters at

402: $\mathrm{E}$, is equally distributed among $\mathrm{C}_{1,2}$, thus

403: $I_{\mathrm{C}_1}=I_{\mathrm{C}_2}=-I_\mathrm{E}/2$.

404: %

405: Since this structure in Fig.~\ref{fig:switch1} is essentially two-dimensional,

406: we have to take also the polarisation of the laser field into account.  We

407: assume it to be linear with an polarisation angle $\phi$ as sketched in the

408: inset of Fig.~\ref{fig:switch1}.  The effective driving amplitudes of the

409: orbitals which are attached to the leads acquire now a geometric factor which

410: is only the same for both orbitals $\mathrm{C}_1$ and $\mathrm{C}_2$ when

411: $\phi=0$.  For any other polarisation angle, the mentioned symmetry is broken

412: and the outgoing currents may be different from each other.  The difference may

413: be huge, as depicted in Fig.~\ref{fig:switch1}.  Their ratio varies from unity

414: for $\phi=0$ up to the order of 100 for $\phi=60^\circ$.  Thus, adapting the

415: polarisation angle enables one to route the current towards the one or the other

416: drain.

417: %%%% Figure 3 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

418: \begin{figure}

419:   \begin{center}

420:     \includegraphics[width=7.5truecm]{fig3.eps}

421:   \end{center}

422:   \caption{Average currents through contacts $\mathrm{C}_1$ (solid)

423:     and $\mathrm{C}_2$ (broken) as a function of the polarisation angle $\phi$

424:     for the three-terminal device depicted in the inset.  The chemical

425:     potentials are $\mu_\mathrm{E}=-\mu_{\mathrm{C}_1} =-\mu_{\mathrm{C}_2}=50

426:     \Delta$; the on-site energies $E_n=0$.  The driving field is specified by

427:     the strength $A=25\Delta$ and the angular frequency $\Omega=10\Delta/\hbar$;

428:     the effective coupling is $\hbar\Gamma=0.1\Delta$ and the temperature $k_B

429:     T=0.25\Delta$.  The maximal value of the current ratio

430:     $I_{\mathrm{C}_1}/I_{\mathrm{C}_2}\approx 100$ is assumed at $\phi=60^\circ$. }

431:   \label{fig:switch1}

432: \end{figure}%

433: %%%% Figure 3 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

434: 

435: Alternatively, one can keep the polarisation angle at $\phi=0$ and break the 

436: reflection symmetry by using an intrinsically asymmetric molecule, as

437: sketched in the inset of Fig.~\ref{fig:switch2}. This allows to 

438: control sensitively the ratio of the outgoing currents by the strength $A$ of the

439: external field, cf.\ Fig.~\ref{fig:switch2}.  The switching range comprises

440: up to four orders of magnitude with an exponential sensitivity.

441: %%%% Figure 4 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

442: \begin{figure}

443:   \begin{center}

444:     \includegraphics[width=7.5truecm]{fig4.eps}

445:   \end{center}

446:   \caption{Ratio of the outgoing average currents versus driving strength $A$

447:     for the three-terminal device at a polarisation angle $\phi=0$.

448:     The filled circle in the inset depicts a site with an on-site energy

449:     $E_{\mathrm{C}_1}$ that differs from the others.

450:     All other on-site energies and parameters as in Fig.~\ref{fig:switch1}.

451:     }

452:   \label{fig:switch2}

453: \end{figure}

454: %%%% Figure 4 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

455: 

456: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

457: \section{Concluding remarks}

458: 

459: We have presented a method for the efficient numerical computation of currents

460: through periodically time-dependent networks with two or more contacts to

461: external leads.  The application to two types of setups substantiated that

462: external fields bear a wealth of possibilities for the manipulation of

463: electrical currents: in a molecular wire the current can be suppressed by

464: proper time-dependent fields.  In a three terminal device, it is possible to

465: route by tailored optical fields the current that enters from a source towards

466: the one or the other drain.

467: 

468: The authors hope that their proposals will motivate experimentalists to accept

469: the challenge of implementing the proposed molecular wire schemes in the

470: laboratory.  The two-terminal current gate can possibly be realized using break

471: junctions exposed to a laser field.  Alternatively, one could use a

472: self-assembled, laser-irradiated maze-like layer of sparsely distributed

473: conducting molecules on a conducting surface. Then by positioning a scanning

474: tunnelling microscope tip directly over one such molecule, it should be

475: possible to measure the features of the predicted gating behaviour.

476: Experimentally more ambitious is the realization of the arrangement in

477: Fig.~\ref{fig:switch1} with a planar molecule contacted to three electrodes.

478: Here again, laser-irradiated self-assemblies of molecules such as carbon

479: nanotube complexes or of biomolecules like metalised DNA \cite{Braun1998a}, or

480: the use of cationic lipid-DNA complexes \cite{Raedler1998a} as DNA-nanocables,

481: with the centre-molecule covalently bound to such planar structures, might make

482: the experiment feasible.

483: 

484: A completely different realization of our findings should be possible in

485: semiconductor heterostructures. There, instead of a molecule, coherently

486: coupled quantum dots \cite{Blick1996a} form the central system. Furthermore,

487: owing to the lower level spacings, the suitable frequency of the coherent

488: radiation source is then in the microwave spectral range.

489: 

490: 

491: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

492: \ack

493: We appreciate helpful discussions with Igor Goychuk and Gert-Ludwig Ingold.

494: This work has been supported by the Deutsche Forschungsgemeinschaft

495: through SFB 486  and by the Volkswagenstiftung under grant No.\ I/77 217.

496: 

497: 

498: \begin{thebibliography}{10}

499: \expandafter\ifx\csname url\endcsname\relax

500:   \def\url#1{\texttt{#1}}\fi

501: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi

502: 

503: \bibitem{Reed1997a}

504: M.~A. Reed, C.~Zhou, C.~J. Muller, T.~P. Burgin, J.~M. Tour, Conductance of a

505:   molecular junction, Science 278 (1997) 252.

506: 

507: \bibitem{Cui2001a}

508: X.~D. Cui, A.~Primak, X.~Zarate, J.~Tomfohr, O.~F. Sankey, A.~L. Moore, T.~A.

509:   Moore, D.~Gust, G.~Harris, S.~M. Lindsay, Reproducible measurement of

510:   single-molecule conductivity, Science 294 (2001) 571.

511: 

512: \bibitem{Joachim2000a}

513: C.~Joachim, J.~K. Gimzewski, A.~Aviram, Electronics using hybrid-molecular and

514:   mono-layer devices, Nature 408 (2000) 541.

515: 

516: \bibitem{Schon2001a}

517: J.~H. Sch\"on, H.~Meng, Z.~Bao, Self-assembled monolayer organic field-effect

518:   transistors, Nature 413 (2001) 713.

519: 

520: \bibitem{Reichert2002a}

521: J.~Reichert, R.~Ochs, D.~Beckmann, H.~Weber, M.~Mayor, H.~v.~Loehneysen,

522:   Driving current through single organic molecules, Phys. Rev. Lett. 88 (2002)

523:   176804.

524: 

525: \bibitem{Nitzan2001a}

526: A.~Nitzan, Electron transmission through molecules and molecular interfaces,

527:   Ann. Rev. Phys. Chem. 52 (2001) 681.

528: 

529: \bibitem{Mujica1994a}

530: V.~Mujica, M.~Kemp, M.~A. Ratner, Electron conduction in molecular wires. i. a

531:   scattering formalism, J. Chem. Phys. 101 (1994) 6849.

532: 

533: \bibitem{Nitzan2001b}

534: A.~Nitzan, A relationship between electron-transfer rates and molecular

535:   conduction, J. Phys. Chem. A 105 (2001) 2677.

536: 

537: \bibitem{Petrov2001a}

538: E.~G. Petrov, P.~H\"anggi, Nonlinear electron current through a short molecular

539:   wire, Phys. Rev. Lett. 86 (2001) 2862.

540: 

541: \bibitem{Heurich2001a}

542: J.~Heurich, J.~C. Cuevas, W.~Wenzel, G.~Sch\"on, Electrical transport through

543:   single-molecule junctions: from molecular orbitals to conduction channels,

544:   cond-mat/  (2001) 0110147.

545: 

546: \bibitem{Grossmann1991a}

547: F.~Grossmann, T.~Dittrich, P.~Jung, P.~H\"anggi, Coherent destruction of

548:   tunneling, Phys. Rev. Lett. 67 (1991) 516.

549: 

550: \bibitem{Grossmann1992a}

551: F.~Gro{\ss}mann, P.~H\"anggi, Localization in a driven two-level dynamics,

552:   Europhys. Lett. 18 (1992) 571.

553: 

554: \bibitem{Morillo1993a}

555: M.~Morillo, R.~I. Cukier, Control of proton transfer reactions with external

556:   fields, J. Chem. Phys. 98 (1993) 4548.

557: 

558: \bibitem{Goychuk1996a}

559: I.~A. Goychuk, E.~G. Petrov, V.~May, Control of the dynamics of a dissipative

560:   two-level system by a strong periodic field, Chem. Phys. Lett. 253 (1996)

561:   428.

562: 

563: \bibitem{Grifoni1998a}

564: M.~Grifoni, P.~H\"anggi, Driven quantum tunneling, Phys. Rep. 304 (1998) 229.

565: 

566: \bibitem{Shirley1965a}

567: J.~H. Shirley, Solution of the {S}chr\"odinger equation with a {H}amiltonian

568:   periodic in time, Phys. Rev. 138 (1965) B979.

569: 

570: \bibitem{Sambe1973a}

571: H.~Sambe, Steady states and quasienergies of a quantum-mechanical system in an

572:   oscillating field, Phys. Rev. A 7 (1973) 2203.

573: 

574: \bibitem{Lehmann2002b}

575: J.~Lehmann, S.~Kohler, P.~H\"anggi, A.~Nitzan, Molecular wires acting as

576:   quantum ratchets, Phys. Rev. Lett. 88 (2002) 228305.

577: 

578: \bibitem{Blumel1991a}

579: R.~Bl\"umel, A.~Buchleitner, R.~Graham, L.~Sirko, U.~Smilansky, H.~Walter,

580:   Dynamical localisation in the microwave interaction of {R}ydberg atoms: The

581:   influence of noise, Phys. Rev. A 44 (1991) 4521.

582: 

583: \bibitem{Bruder1994a}

584: C.~Bruder, H.~Schoeller, Charging effects in ultrasmall quantum dots in the

585:   presence of time-varying fields, Phys. Rev. Lett. 72 (1994) 1076.

586: 

587: \bibitem{Demming1998a}

588: F.~Demming, J.~Jersch, K.~Dickmann, P.~I. Geshev, Calculation of the field

589:   enhancement on laser-illuminated scanning probe tips by the boundary element

590:   method, Appl. Phys. B 66 (1998) 593.

591: 

592: \bibitem{Mujica2000a}

593: V.~Mujica, A.~E. Roitberg, M.~Ratner, Molecular wire conductance: Electrostatic

594:   potential spatial profile, J. Chem. Phys. 112 (2000) 6834.

595: 

596: \bibitem{Chen1999a}

597: J.~Chen, M.~A. Reed, A.~M. Rawlett, J.~M. Tour, Large on-off ratios and

598:   negative differential resistance in a molecular electronic device, Science

599:   286 (1999) 1550.

600: 

601: \bibitem{Collier2000a}

602: C.~P. Collier, G.~Mattersteig, E.~W. Wong, Y.~Luo, K.~Beverly, J.~Sampaio,

603:   F.~M. Raymo, J.~F. Stoddart, J.~R. Heath, A [2]catenane-based solid state

604:   electronically reconfigurable switch, Science 289 (2000) 1172.

605: 

606: \bibitem{Braun1998a}

607: E.~Braun, Y.~Eichen, U.~Sivan, G.~Ben-Yoseph, {DNA}-templated assembly and

608:   electrode attachment of a conducting silver wire, Nature 391 (1998) 775.

609: 

610: \bibitem{Raedler1998a}

611: O.~J. R\"adler, I.~Koltover, A.~Jamieson, T.~Salditt, C.~R. Safinya, Structure

612:   and interfacial aspects of self-assembled cationic lipid-{DNA} gene carrier

613:   complexes, Langmuir 14 (1998) 4272.

614: 

615: \bibitem{Blick1996a}

616: R.~H. Blick, R.~J. Haug, J.~Weis, D.~Pfannkuche, K.~v. Klitzing, K.~Eberl,

617:   Single-electron tunneling through a double quantum dot: The artificial

618:   molecule, Phys. Rev. B 53 (1996) 7899.

619: 

620: \end{thebibliography}

621: 

622: 

623: \end{document}

624: