1: % This manuscript requires REVTeX4

2: %

3: % http://publish.aps.org/revtex4/

4: %

5: \documentclass[%

6: 10pt,%

7: aps,%

8: prb,

9: letterpaper,

10: twocolumn,%preprint,%

11: superscriptaddress,%

12: preprintnumbers,%

13: floatfix%

14: ]{revtex4}

15: 

16: \usepackage[usenames]{color}

17: \usepackage{times}

18: \usepackage{dcolumn}% Align table columns on decimal point

19: \usepackage{bm}% bold math

20: \usepackage{ifthen}

21: \usepackage{amssymb}

22: \usepackage{amsfonts}

23: \usepackage{amsmath}

24: \usepackage[dvips]{graphicx}

25: 

26: 

27: \def\figurename{\textbf{\textsf{Figure}}}

28: \renewcommand \thefigure {\textbf{\textsf{\arabic{figure}}}}

29: 

30: \newcommand\myCaption[2]{\begin{flushleft}\fontsize{7}{10}\selectfont\refstepcounter{figure}%

31: \noindent\textsf{\figurename}~\fontsize{7}{10}\selectfont\thefigure$\vert$~\fontsize{7}{10}\selectfont{\textsf{\textbf{#1}}}

32: {\textbf {#2}}\end{flushleft}\small\selectfont}

33: 

34: \setlength{\parindent}{3.truemm}

35: 

36: \def\ee{{\rm e}}

37: \def\ii{\textrm{i}\,\!}

38: 

39: \newcommand{\pippo}[1]{\centerline{#1}}

40: \newcommand{\beann} {\begin{eqnarray*}}

41: \newcommand{\eeann} {\end{eqnarray*}}

42: \newcommand{\bea} {\begin{eqnarray}}

43: \newcommand{\eea} {\end{eqnarray}}

44: \newcommand{\labs} {\left\vert}

45: \newcommand{\rabs} {\right\vert}

46: \newcommand{\lsb} {\left[}

47: \newcommand{\rsb} {\right]}

48: \newcommand{\lrb} {\left(}

49: \newcommand{\rrb} {\right)}

50: \newcommand{\lcb} {\left\{}

51: \newcommand{\rcb} {\right\}}

52: \newcommand{\lab} {\left\langle}

53: \newcommand{\rab} {\right\rangle}

54: \newcommand{\ve} {\varepsilon}

55: \newcommand{\com} {\%}

56: \newcommand{\postit}[1]{\textcolor{red}{[[ \textbf{#1} ]]}}

57: \newcommand{\vect} {\mathbf}

58: 

59: \newcommand{\ReP} {{\rm Re}~}

60: \newcommand{\ImP} {{\rm Im}~}

61: \newcommand{\Ham} {{\mathcal H}}

62: \newcommand{\Gf} {{\mathcal G}}

63: \newcommand{\Glead} {g}

64: \newcommand{\T} {{\mathcal T}}

65: \newcommand{\Cond} {G}

66: \newcommand{\gc}{\,^\circ\mathrm{C}}

67: \newcommand{\g}{^\circ}

68: \newcommand{\ie}{\textit{i.e.~}}

69: \newcommand{\eg}{\textit{e.g.~}}

70: \newcommand{\lnothing} {\left.}

71: \newcommand{\rnothing} {\right.}

72: %\newcommand{\appendixname} {Supplementary Information}

73: 

74: \begin{document}

75: \date{\today}

76: 

77: \title{\huge\bfseries\noindent\sloppy \textsf{Tuning the conductance of a molecular switch}}

78: 

79: \author{\textsf{Miriam del Valle}}

80: \affiliation{Institute for Theoretical Physics, University of

81: Regensburg, D-93040 Regensburg, Germany} \affiliation{Departamento

82: de F\'{\i}sica Te\'{o}rica de la Materia Condensada, Universidad

83: Aut\'{o}noma de Madrid, E-28049 Madrid, Spain}

84: \author{\textsf{Rafael Guti{\'e}rrez}}

85: \affiliation{Institute for Theoretical Physics, University of

86: Regensburg, D-93040 Regensburg, Germany}

87: \author{\textsf{Carlos Tejedor}}

88: \affiliation{Departamento de F\'{\i}sica Te\'{o}rica de la Materia

89: Condensada, Universidad Aut\'{o}noma de Madrid, E-28049 Madrid,

90: Spain}

91: \author{\textsf{Gianaurelio Cuniberti\footnote{Corresponding author}}}

92: \affiliation{Institute for Theoretical Physics, University of

93: Regensburg, D-93040 Regensburg, Germany}

94: 

95: \maketitle \small \noindent \textbf{ The ability to control the

96: conductance of single molecules will have a major impact in

97: nanoscale

98: electronics\cite{joachim05,lindsay06,ralph05,Donhauser01,Dujardin05,Wu04,ElbingOKFHWEWM05,cavin06,choi06,MorescoMRTGJ01,dulic03}.

99: Azobenzene, a molecule that changes conformation as a result of a

100: \textit{trans}/\textit{cis} transition when exposed to radiation,

101: could form the basis of a light-driven molecular

102: switch\cite{HugelHCMSG02,ZhangDCZRK04,Ratner06}. It is therefore

103: crucial to clarify the electrical transport characteristics of this

104: molecule. Here, we investigate theoretically charge transport in a

105: system in which a single azobenzene molecule is attached to two

106: carbon nanotubes. In clear contrast to gold electrodes, the

107: nanotubes can act as true nanoscale electrodes and we show that the

108: low-energy conduction properties of the junction may be dramatically

109: modified by changing the topology of the contacts between the

110: nanotubes and the molecules, and/or the chirality of the nanotubes

111: (that is, zigzag or armchair). We propose experiments to demonstrate

112: controlled electrical switching with nanotube electrodes.}

113: 

114: 

115: Recent theoretical studies\cite{ZhangDCZRK04,Ratner06} have explored

116: the use of azobenzene as an electronic switch when contacted by Au

117: electrodes. Promising results were obtained, as a significant change

118: in conductance was seen between the two conformations. Although a

119: mechanism was proposed that allowed for a certain movement of the

120: Au-leads, azobenzene and similar molecules undergoing isomerization

121: transitions require more versatile nanocontacts that adjust

122: appropriately to the change of length in going from one isomeric

123: state to the other. Moreover,

124:  the coupling to Au electrodes only leads

125: to a broadening of the molecular orbitals, without essentially

126: perturbing the intrinsic molecular electronic structure. Although

127: this may be a positive feature in some situations, more exciting is

128: the possibility of having nanoscale electrodes with electronic

129: properties  that may lead, in contrast to Au electrodes, to a strong

130: modification of the low-energy molecular electronic structure and

131: hence to dramatic changes in conductance. The most attractive

132: candidates are carbon nanotubes (CNTs)\cite{GuoSKWPTHCHOYBWHKN06},

133: which are essentially one-dimensional systems whose conduction

134: character (metallic or semiconducting) can be easily tuned by

135: changing their chirality.

136: 

137: We focus in this paper  on the transport characteristics of single

138: \textit{trans} and \textit{cis} azobenzene contacted by metallic

139: CNTs with two typical chiralities: armchair and zigzag. In order to

140: improve the molecule-electrode coupling, the hydrogen atoms at the

141: para-positions on the molecules are substituted by NHCO-groups,

142: which have been used as linkers to CNT electrodes in recent

143: transport experiments (A. Holleitner, private communication). Our

144: main aims are to highlight the dependence of charge transport on the

145: electronic structure of the nanotube electrodes, as well as the

146: influence of the linker groups, which are known to play an important

147: role\cite{steigerwald06}. It turns out that the oxygen atom plays a

148: critical role in determining the conductance near the Fermi level.

149: Finally, the sensitivity in the electrical response of the isomers

150: may be exploited to identify them using electrical transport

151: measurements as well as to realize a molecular switch.

152: 

153: %------------------------------------------------------------------------%

154: \begin{figure}[b]

155: \pippo{\includegraphics[width=.999\linewidth]{Figure_1.eps}}

156: \myCaption{\label{fig:isolated} Low-energy electronic states of

157: azobenzene including the NHCO linkers.}{The positions of the

158: frontier molecular orbitals of the \textit{trans} (left) and

159: \textit{cis} (right) isomers are shown. The HOMO-LUMO gap is

160: slightly smaller in the \textit{trans} state, which is also the

161: lowest-energy configuration. The presence of the linkers does not

162: appreciably modify the geometric structure of the isomers. Also

163: shown is a density isosurface plot for the HOMO and LUMO orbitals.

164: The \textit{trans}-HOMO is mainly localized around the N-N dimer

165: with almost no weight on the linker groups. Notice that an extra

166: hydrogen saturates the linker. On attaching the molecules to the CNT

167: electrodes, this H-atom is removed and plays no further role.}

168: \end{figure}

169: %------------------------------------------------------------------------%

170: 

171: The \textit{trans} isomer is characterized by a planar structure,

172: but in the \textit{cis} isomer the benzene aromatic rings are tilted

173: with respect to each other, as seen in Fig.~\ref{fig:isolated}.

174: Total energy calculations yield an energy difference between the

175: isomers of $\sim 0.2\ e\textrm{V}$, the \textit{trans} state having

176: the lowest energy and smallest highest occupied molecular orbital

177: (HOMO)/lowest unoccupied molecular orbital (LUMO) gap of  $\sim

178: 1.98\ e\textrm{V}$. On addition of the linkers, the gap slightly

179: increases to $ 2.03 \ e\textrm{V}$. Conversely, the gap for the

180: \textit{cis} isomer decreases from $ 2.55 \ e\textrm{V}$ to  $ 2.34

181: \ e\textrm{V}$. However, the isomer geometries, when the linkers are

182: added, are not appreciably modified after relaxation. In

183: Fig.~\ref{fig:isolated} we show the energetic position of the

184: isomers' frontier orbitals  as well as the optimized geometries.

185: 

186: 

187: %------------------------------------------------------------------------%

188: \begin{figure}[t]

189: \pippo{\includegraphics[width=.999\linewidth]{Figure_2.eps}}

190: \myCaption{\label{fig:55CNT} Linear conductance at zero bias for a

191: (5,5) armchair-\textit{trans}(\textit{cis})-armchair junction.}{As a

192: reference we indicate by vertical lines the  positions of the HOMO

193: and LUMO orbitals of the isolated \textit{trans} (blue dashed lines)

194: and \textit{cis} (red dotted lines) molecules. All energies are

195: shifted to set the Fermi energy at $ 0 \ e\textrm{V}$, as only

196: energy differences from the Fermi level are revealed in transport

197: observables. Top panel: density isosurface plot of a selected

198: molecular orbital of the extended \textit{trans} molecule (at an

199: energy indicated in the conductance plot by the arrow near the Fermi

200: level). The low weight on the molecule reflects the rather low

201: conductance at this energy. The inset is an enlargement of the

202: CNT-molecule interface showing the formation of a carbon hexagon

203: with an oxygen defect on relaxation. This configuration is the same

204: for both isomers. Note that the carbon atom denoted by C$_{1}$ forms

205: a "bottleneck" in the junction, which means that the conductance is

206: largely determined by the spectral weight on the O-C$_1$ bond.}

207: \end{figure}

208: %------------------------------------------------------------------------%

209: 

210: Let us consider  the CNT-azobenzene-CNT molecular junction. The

211: complexity of the system leads to a very structured potential-energy

212: hypersurface with  many metastable states; for the sake of

213: simplicity we only consider coaxial arrangements of the left and

214: right CNTs. Different junction geometries varying in the way the

215: molecule is attached to the CNT surfaces were then optimized using

216: conjugated-gradient techniques, and the lowest-energy configurations

217: were used for the transport calculations. We note at this point that

218: the energy differences between the  investigated metastable states

219: are roughly of the order of a few $\textrm{m}e\textrm{V}$ per atom,

220: so that our transport calculations can be considered as providing

221: expected trends. Other configurations close in energy may lead to

222: slightly different quantitative results. However, the basic

223: conclusions of this paper should not be affected.

224: 

225: %------------------------------------------------------------------------%

226: \begin{figure}[t]

227: \pippo{\includegraphics[width=.999\linewidth]{Figure_3.eps}}

228: \myCaption{\label{fig:90CNT} Linear conductance at zero bias for a

229: (9,0) zigzag-\textit{trans}(\textit{cis})-zigzag junction.}{As in

230: Fig.~\ref{fig:55CNT}, the vertical dashed (dotted) lines indicate

231: the positions of the frontier orbitals of the isolated

232: \textit{trans}(\textit{cis}) isomers. Notice the increased

233: conductance around the Fermi energy for both isomers as well as the

234: more complex structure of the spectrum in this energy region. This

235: is closely related to the strongly localized edge state at the Fermi

236: level in zigzag tubes. Top panel: density isosurface plot of a

237: selected orbital of the extended \textit{trans} molecule (at an

238: energy indicated by an arrow in the conductance plot). Note the much

239: higher weight on the molecule when comparing with the armchair

240: junction in Fig.~\ref{fig:55CNT}, which explains the higher

241: transmission at this energy. The inset shows the local atomic

242: structure at the CNT-molecule interface, where a pentagon is formed.

243: This local bonding topology is similar for both isomers.}

244: \end{figure}

245: %------------------------------------------------------------------------%

246: 

247: 

248: Concerning the CNT-molecule contact, two direct covalent bonds are

249: formed between the azobenzene complex and the CNT leads: a C$_{\rm

250: CNT}$-C$_{1}$ bond and a C$_{\rm CNT}$-O bond. C$_{1}$ denotes the

251: carbon atom belonging to the linker. The bond topology is, however,

252: different for armchair and zigzag tubes, as shown in  the insets of

253: Figs.~\ref{fig:55CNT} and \ref{fig:90CNT}. In the former case a

254: hexagon is formed, and in the latter case a pentagon is built. This

255: appears to be a common feature of the bonding topology, as other

256: initial configurations led to similar connectivities upon

257: relaxation. In the \textit{trans} configuration attached to armchair

258: leads, one of the benzene rings slightly rotates out of plane by

259: about $8\g$ so that the relaxed state deviates from the planar

260: geometry found in the isolated molecule. This structural distortion

261: becomes even stronger when attaching the \textit{trans} molecule to

262: zigzag tubes ($14\g$). On the other hand, for the \textit{cis}

263: isomer, geometrical modifications on attachment to the leads do not

264: depend so much on the chirality of the CNTs, and consist essentially

265: of a slight reduction of the tilt angle.

266: 

267: Once the geometries were optimized, we first focused on the

268: dependence of the linear conductance on the charge injection energy

269: (Figs.~\ref{fig:55CNT} and \ref{fig:90CNT}) for armchair and zigzag

270: junctions, respectively. It turns out that the \textit{trans}

271: configuration shows an overall better ``transparency'' (higher

272: conductance)  than the \textit{cis} state, independently of the CNT

273: chirality, thus suggesting that the switching behaviour may already

274: be expected at low bias. The conductance of the

275: armchair-\textit{cis} junction displays a parabola-like gap around

276: the Fermi level, indicating that transport can only take place

277: through tunnelling, and thus very low currents are expected.

278: However, several resonances are seen for the \textit{trans} state

279: around the Fermi energy, which, despite their relative low

280: intensity, may give rise to resonant transport. These features arise

281: from the hybridization of surface states with molecular orbitals.

282: The resulting mixed states have a non-negligible overlap with the

283: CNT extended bulk states and thus can contribute to transport. The

284: fact that such states lie within a low-energy window around $E_{\rm

285: F}$ can make conductance manipulation more feasible because these

286: states will sensitively depend on the molecule-CNT contact topology,

287: as can be seen from our calculations. The low transmission of the

288: \textit{cis}-state is related to its distorted geometry, which

289: considerably breaks the $\pi$-conjugation along the molecular frame.

290: 

291: Turning now to the zigzag-azobenzene junctions, the conductance

292: appears to be considerably larger for both isomers than for armchair

293: junctions, as seen in Fig.~\ref{fig:90CNT}. One should notice at

294: this point that for both chiralities there are essentially two

295: interfering pathways  from the CNT to the azobenzene molecule, which

296: merge in the carbon atom belonging to the linker. Our calculations

297: show that the O-C$_{1}$ pathway completely dominates the charge

298: transport efficiency through the junction, the oxygen atom in the

299: NHCO linker hereby playing a crucial role. Because of its higher

300: electroaffinity, the oxygen tends to deplete the spectral weight on

301: the C$_{1}$ atom and thus acts as a chemical gate at the atomic

302: scale. This mechanism is very effective for armchair junctions, but

303: fails to have a dominant effect for zigzag junctions. Our

304: density-functional-theory-based calculations as well as a

305: $\pi$-orbital model hamiltonian show that, contrary to the armchair

306: system, in zigzag junctions a much larger spectral weight on the

307: C$_{1}$ atom persists for energies around the Fermi level. The

308: reason for this is that the surface density of states of a zigzag

309: tube has, as is well known, a large spectral weight close to the

310: Fermi energy (corresponding to a localized state at the zigzag

311: edge)\cite{Fukui05}. Moreover, the strong edge resonance manifests

312: itself in the electrode self-energies $\Sigma_{\rm L/R}(E)$ and

313: leads to a rather complex modification of the lower-lying molecular

314:  electronic states which is absent for armchair junctions.

315: Thus, the conductance difference for both junctions  is a result of

316: the interplay between local (interface) chemistry and electrode

317: surface electronic structure.

318: 

319: 

320: %------------------------------------------------------------------------%

321: \begin{figure}[t]

322: \pippo{\includegraphics[width=.999\linewidth]{Figure_4.eps}}

323: \myCaption{\label{fig:IV} Average conductance and current-voltage

324: characteristics of the CNT-azobenzene junctions.}{Three possible

325: CNT-molecule junctions are shown:

326: zigzag(ZZ)-\textit{trans/cis}-zigzag,

327: armchair(AC)-\textit{trans/cis}-zigzag, and

328: armchair-\textit{trans/cis}-armchair. A systematic hierarchy can be

329: seen, showing that the \textit{trans} state is always better

330: conducting than the \textit{cis} one and that armchair CNTs are

331: always detrimental to junction conductivity. }

332: \end{figure}

333: %------------------------------------------------------------------------%

334: 

335: 

336: To round off our discussion, we have calculated the current-voltage

337: ($I$-$V$) characteristics of the junctions to confirm the possible

338: switching behaviour already suggested by the analysis of the linear

339: conductance. To reveal in a clearer way the effects of chirality on

340: transport, a hybrid armchair-azobenzene-zigzag system was also

341: investigated. The results for the three cases are displayed in

342: Fig.~\ref{fig:IV}. Charge transport for the \textit{cis} state is

343: strongly suppressed already at low bias for all junctions,

344: indicating a switching behaviour. Note that the

345: zigzag-\textit{trans}-zigzag junction shows up to three orders of

346: magnitude larger currents than the other two junctions, where

347: armchair tubes were used. This is even more dramatically seen in the

348: left panel of Fig.~\ref{fig:IV} where the average conductance $I/V$

349: is shown. These results further support the conclusion that the

350: electronic structure of both molecule and electrodes is crucial in

351: determining the low-voltage transport when nanoscale electrodes are

352: used. Another remarkable point is the strong rectifying behaviour

353: found in the hybrid armchair-\textit{trans}-zigzag junction. The

354: obvious chirality dependence of the effect suggests a different

355: mechanism to realize diode-like functions, which would not be

356: directly related to intrinsic electronic structure asymmetries of

357: the molecule as in the paradigmatic case of the Aviram-Ratner

358: molecular rectifier\cite{AviramR74}. We would further like to

359: mention that our equilibrium treatment of transport is validated by

360: recent non-equilibrium calculations\cite{Ratner06} using Au

361: electrodes which showed a similar qualitative behaviour for the two

362: isomers, that is,~$I_{\textit{trans}}(V)\gg I_{\textit{cis}}(V)$.

363: 

364: Finally, we would like to comment on possible experimental

365: realizations of the conformational switching when using CNT

366: electrodes. One precondition to achieve it is to increase the

367: flexibility of the junction, either by using longer linkers (which

368: should not, however, be detrimental to conduction by inducing, for

369: example, tunnel barriers) or by enlarging the electrode mobility.

370: One should keep in mind that the size of the spatial gap where the

371: molecule is placed is considerably smaller than the length of the

372: CNTs. Thus, small lateral variations in the position of the CNTs can

373: lead to a large change in the inter-electrode spacing. Another

374: possibility would lie in using telescopic CNTs as electrodes and

375: attaching the molecule to one of the inner shells\cite{zettl00}.

376: Additionally, we might expect that for a kinematically constrained

377: molecule such as the one studied here, the ``standard''

378: isomerization pathway would in principle be difficult to realize.

379: Although this might be true for bulk Au electrodes, the

380: one-dimensional nature of CNTs can considerably increase the

381: mechanical flexibility of the junction. Alternative pathways through

382: other metastable states compatible with the constrains can therefore

383: become effective. Indeed, several of the relaxed junctions displayed

384: geometric characteristics lying between those of the \textit{trans}

385: and \textit{cis} states. Light irradiation or mechanical actions

386: could trigger the transition from these metastable configurations to

387: the final \textit{cis} state. Another advantage of CNT electrodes

388: lies in the fact that they may posses a negative thermal expansion

389: factor\cite{Tomanek04}. A combination with other metallic electrodes

390: may thus lead to thermally more stable junctions, because normal

391: metal electrodes would expand upon heating and hence reduce the

392: junction flexibility.

393: 

394: \medskip%%

395: 

396: \fontsize{7.5}{9}\selectfont \noindent \textsf{\textbf{METHODS}}\\

397: We have used a density-functional-tight-binding (DFTB)

398: methodology\cite{FrauenheimSEHJPSS00} combined with Green function techniques\cite{PecchiaC04}

399: for geometry optimization and  transport calculations

400:  in the frame of the Landauer approach. The linear conductance $G$ can be obtained from the

401: quantum mechanical transmission probability at zero voltage:

402: $G=G_{0}T(E,V=0)$ (where $G_{0}=2e^2/h$ is the conductance quantum,

403: $e$ is the electron charge and $h$ is the Planck constant). The

404: transmission probability is related to the molecular Green function

405: $g(E)$ and the electrodes' self-energies $\Sigma_{\rm{L/R}}(E)$ by

406: $T(E,V)=4{\rm{Tr}}[g^{\dagger}(E) \ImP \Sigma_{\rm{R}}(E) g(E) \ImP

407: \Sigma_{\rm{L}}(E)]$ (ref.\onlinecite{FisherL81}). We stress that

408: when calculating the current $I(V)$, the full voltage-dependent

409: $T(E,V)$ is  used: $I(V)=(2e/h) \, \int dE \, \left[ f_{\rm

410: L}(E)-f_{\rm R}(E) \right] T(E,V) $. In the last expression

411: $f_{\rm{L/R}}(E)$ are Fermi functions of the left and right

412: electrodes.  $T(E,V)$ is calculated for the so called extended

413: molecule, which contains the azobenzene molecule, the linkers, and

414: the electrodes' surface atoms. This allows the molecule-electrode

415: coupling and the molecule electronic structure to be treated on the

416: same footing\cite{GrossmannGS02a,PecchiaC04}.

417: 

418: \fontsize{7}{10}\selectfont\sf%

419: \begin{thebibliography}{10}

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

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

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

423: \providecommand{\bibinfo}[2]{#2}

424: \providecommand{\eprint}[2][]{\url{#2}}

425: 

426: 

427: \bibitem{joachim05}

428: \bibinfo{author}{Joachim, C.} \& \bibinfo{author}{Ratner, M.~A.}

429: \newblock \bibinfo{title}{Molecular electronics: Some views on transport

430:   junctions and beyond}.

431: \newblock \emph{\bibinfo{journal}{Proc.\ Natl.\ Acad.\ Sci.\ USA}}

432:   \textbf{\bibinfo{volume}{102}}, \bibinfo{pages}{8801--8808}

433:   (\bibinfo{year}{2005}).

434: 

435: \bibitem{lindsay06}

436: \bibinfo{author}{Visoly-Fisher, I.} \emph{et~al.}

437: \newblock \bibinfo{title}{Conductance of a biomolecular wire}.

438: \newblock \emph{\bibinfo{journal}{Proc.\ Natl.\ Acad.\ Sci.\ USA}}

439:   \textbf{\bibinfo{volume}{103}}, \bibinfo{pages}{8686--8690}

440:   (\bibinfo{year}{2006}).

441: 

442: \bibitem{ralph05}

443: \bibinfo{author}{Champagne, A.~R.}, \bibinfo{author}{Pasupathy, A.~N.} \&

444:   \bibinfo{author}{Ralph, D.~C.}

445: \newblock \bibinfo{title}{Mechanically adjustable and electrically gated

446:   single-molecule transistors}.

447: \newblock \emph{\bibinfo{journal}{Nano Lett.}} \textbf{\bibinfo{volume}{5}},

448:   \bibinfo{pages}{305--308} (\bibinfo{year}{2005}).

449: 

450: \bibitem{Donhauser01}

451: \bibinfo{author}{Donhauser, Z.~J.} \emph{et~al.}

452: \newblock \bibinfo{title}{Conductance switching in single molecules through

453:   conformational changes}.

454: \newblock \emph{\bibinfo{journal}{Science}} \textbf{\bibinfo{volume}{292}},

455:   \bibinfo{pages}{2303--2307} (\bibinfo{year}{2001}).

456: 

457: \bibitem{Dujardin05}

458: \bibinfo{author}{Lastapis, M.} \emph{et~al.}

459: \newblock \bibinfo{title}{Picometer-scale electronic control of molecular

460:   dynamics inside a single molecule}.

461: \newblock \emph{\bibinfo{journal}{Science}} \textbf{\bibinfo{volume}{308}},

462:   \bibinfo{pages}{1000--1003} (\bibinfo{year}{2005}).

463: 

464: \bibitem{Wu04}

465: \bibinfo{author}{Wu, S.~W.}, \bibinfo{author}{Nazin, G.},

466:   \bibinfo{author}{Chen, X.}, \bibinfo{author}{Qiu, X.~H.} \&

467:   \bibinfo{author}{Ho, W.}

468: \newblock \bibinfo{title}{Control of relative tunneling rates in single

469:   molecule bipolar electron transport}.

470: \newblock \emph{\bibinfo{journal}{Phys.\ Rev.\ Lett.}}

471:   \textbf{\bibinfo{volume}{93}}, \bibinfo{pages}{236802}

472:   (\bibinfo{year}{2004}).

473: 

474: \bibitem{ElbingOKFHWEWM05}

475: \bibinfo{author}{Elbing, M.} \emph{et~al.}

476: \newblock \bibinfo{title}{A single-molecule diode}.

477: \newblock \emph{\bibinfo{journal}{Proc.\ Natl.\ Acad.\ Sci.\ USA}}

478:   \textbf{\bibinfo{volume}{102}}, \bibinfo{pages}{8815--8820}

479:   (\bibinfo{year}{2005}).

480: 

481: \bibitem{cavin06}

482: \bibinfo{author}{Zhirnov, V.~V.} \& \bibinfo{author}{Cavin, R.~K.}

483: \newblock \bibinfo{title}{Molecular electronics: Chemistry of molecules or

484:   physics of contacts?}

485: \newblock \emph{\bibinfo{journal}{Nature Mater.}}

486:   \textbf{\bibinfo{volume}{5}}, \bibinfo{pages}{11--12} (\bibinfo{year}{2006}).

487: 

488: \bibitem{choi06}

489: \bibinfo{author}{Choi B.-Y.} \emph{et~al.}

490: \newblock \bibinfo{title}{Conformational molecular switch of the azobenzene molecule:

491: A scanning tunneling microscopy study}.

492: \newblock \emph{\bibinfo{journal}{Phys. Rev. Lett.}}

493:   \textbf{\bibinfo{volume}{96}}, \bibinfo{pages}{156106} (\bibinfo{year}{2006}).

494: 

495: \bibitem{MorescoMRTGJ01}

496: \bibinfo{author}{Moresco, F.} \emph{et~al.}

497: \newblock \bibinfo{title}{Conformational changes of single molecules induced by

498:   scanning tunneling microscopy manipulation: A route to molecular switching}.

499: \newblock \emph{\bibinfo{journal}{Phys.\ Rev.\ Lett.}}

500:   \textbf{\bibinfo{volume}{86}}, \bibinfo{pages}{672--675}

501:   (\bibinfo{year}{2001}).

502: 

503: \bibitem{dulic03}

504: \bibinfo{author}{Duli\'c, D.} \emph{et~al.}

505: \newblock \bibinfo{title}{One-way optoelectronic switching of photochromic molecules on gold}.

506: \newblock \emph{\bibinfo{journal}{Phys.\ Rev.\ Lett.}}

507:   \textbf{\bibinfo{volume}{91}}, \bibinfo{pages}{207402}

508:   (\bibinfo{year}{2003}).

509: 

510: \bibitem{HugelHCMSG02}

511: \bibinfo{author}{Hugel, T.} \emph{et~al.}

512: \newblock \bibinfo{title}{Single-molecule optomechanical cycle}.

513: \newblock \emph{\bibinfo{journal}{Science}} \textbf{\bibinfo{volume}{296}},

514:   \bibinfo{pages}{1103--1106} (\bibinfo{year}{2002}).

515: 

516: \bibitem{ZhangDCZRK04}

517: \bibinfo{author}{Zhang, C.} \emph{et~al.}

518: \newblock \bibinfo{title}{Coherent electron transport through an azobenzene

519:   molecule: A light-driven molecular switch}.

520: \newblock \emph{\bibinfo{journal}{Phys.\ Rev.\ Lett.}}

521:   \textbf{\bibinfo{volume}{92}}, \bibinfo{pages}{158301}

522:   (\bibinfo{year}{2004}).

523: 

524: \bibitem{Ratner06}

525: \bibinfo{author}{Zhang, C.} \emph{et~al.}

526: \newblock \bibinfo{title}{Current-voltage characteristics through a single

527:   light-sensitive molecule}.

528: \newblock \emph{\bibinfo{journal}{Phys.\ Rev.\ B}}

529:   \textbf{\bibinfo{volume}{73}}, \bibinfo{pages}{125445}

530:   (\bibinfo{year}{2006}).

531: 

532: \bibitem{GuoSKWPTHCHOYBWHKN06}

533: \bibinfo{author}{Guo, X.} \emph{et~al.}

534: \newblock \bibinfo{title}{Covalently bridging gaps in single-walled carbon

535:   nanotubes with conducting molecules}.

536: \newblock \emph{\bibinfo{journal}{Science}} \textbf{\bibinfo{volume}{311}},

537:   \bibinfo{pages}{356--359} (\bibinfo{year}{2006}).

538: 

539: \bibitem{steigerwald06}

540: \bibinfo{author}{Venkataraman L.} \emph{et~al.}

541: \newblock \bibinfo{title}{Single-molecule circuits with

542: well-defined molecular conductance}.

543: \newblock \emph{\bibinfo{journal}{Nano Lett.}} \textbf{\bibinfo{volume}{6}},

544:   \bibinfo{pages}{458-462} (\bibinfo{year}{2006}).

545: 

546: \bibitem{Fukui05}

547: \bibinfo{author}{Kobayashi, Y.}, \bibinfo{author}{Fukui, K.},

548:   \bibinfo{author}{Enoki, T.}, \bibinfo{author}{Kusakabe, K.} \&

549:   \bibinfo{author}{Kaburagi, Y.}

550: \newblock \bibinfo{title}{Observation of zigzag and armchair edges of graphite

551:   using scanning tunneling microscopy and spectroscopy}.

552: \newblock \emph{\bibinfo{journal}{Phys.\ Rev.\ B}}

553:   \textbf{\bibinfo{volume}{71}}, \bibinfo{pages}{193406}

554:   (\bibinfo{year}{2005}).

555: 

556: \bibitem{AviramR74}

557: \bibinfo{author}{Aviram, A.} \& \bibinfo{author}{Ratner, M.~A.}

558: \newblock \bibinfo{title}{Molecular rectifiers}.

559: \newblock \emph{\bibinfo{journal}{Chem. \ Phys. \ Lett.}}

560:   \textbf{\bibinfo{volume}{29}}, \bibinfo{pages}{277-280}

561:   (\bibinfo{year}{1974}).

562: 

563: \bibitem{zettl00}

564: \bibinfo{author}{Cumings, J.} \& \bibinfo{author}{ Zettl, A.}

565: \newblock \bibinfo{title}{Low-friction nanoscale linear

566: bearing realized from multiwall carbon nanotubes}.

567: \newblock \emph{\bibinfo{journal}{Science}}

568:   \textbf{\bibinfo{volume}{289}}, \bibinfo{pages}{602--604}

569:   (\bibinfo{year}{2000}).

570: 

571: \bibitem{Tomanek04}

572: \bibinfo{author}{Kwon, Y.-K.}, \bibinfo{author}{Berber, S.} \&

573:   \bibinfo{author}{Tomanek, D.}

574: \newblock \bibinfo{title}{Thermal contraction of carbon fullerenes and

575:   nanotubes}.

576: \newblock \emph{\bibinfo{journal}{Phys.\ Rev.\ Lett.}}

577:   \textbf{\bibinfo{volume}{92}}, \bibinfo{pages}{015901}

578:   (\bibinfo{year}{2004}).

579: 

580: \newpage

581: \bibitem{FrauenheimSEHJPSS00}

582: \bibinfo{author}{Frauenheim, T.} \emph{et~al.}

583: \newblock \bibinfo{title}{A self-consistent charge density-functional based

584:   tight-binding method for predictive materials simulations in physics,

585:   chemistry and biology}.

586: \newblock \emph{\bibinfo{journal}{Phys. Stat. Sol. b}}

587:   \textbf{\bibinfo{volume}{217}}, \bibinfo{pages}{41--62}

588:   (\bibinfo{year}{2000}).

589: 

590: \bibitem{PecchiaC04}

591: \bibinfo{author}{Pecchia, A.} \& \bibinfo{author}{Carlo, A.~D.}

592: \newblock \bibinfo{title}{Atomistic theory of transport in organic and

593:   inorganic nanostructures}.

594: \newblock \emph{\bibinfo{journal}{Rep. Prog. Phys.}}

595:   \textbf{\bibinfo{volume}{67}}, \bibinfo{pages}{1497--1561}

596:   (\bibinfo{year}{2004}).

597: 

598: \bibitem{FisherL81}

599: \bibinfo{author}{Fisher, D.~S.} \& \bibinfo{author}{Lee, P.~A.}

600: \newblock \bibinfo{title}{Relation between conductivity and transmission

601:   matrix}.

602: \newblock \emph{\bibinfo{journal}{Phys.\ Rev.\ B}}

603:   \textbf{\bibinfo{volume}{23}}, \bibinfo{pages}{6851--6854}

604:   (\bibinfo{year}{1981}).

605: 

606: \bibitem{GrossmannGS02a}

607: \bibinfo{author}{Cuniberti, G.}, \bibinfo{author}{Grossmann, F.} \&

608:   \bibinfo{author}{Guti\'errez, R.}

609: \newblock \bibinfo{title}{The role of contacts in molecular electronics}.

610: \newblock \emph{\bibinfo{journal}{Advances in Solid State Physics}}

611: \textbf{\bibinfo{volume}{42}},

612:   \bibinfo{pages}{133--149} (\bibinfo{year}{2002}).

613: 

614: \end{thebibliography}

615: 

616: 

617: 

618: %% Define the addendum environment for Supplementary Info, Acknowledgements, etc.

619: \newenvironment{addendum}{%

620:  \setlength{\parindent}{0in}%

621: \fontsize{7}{10}\selectfont\sf%

622:  \begin{list}{\textsf{Acknowledgements}}{%

623:  \setlength{\leftmargin}{0in}%

624:  \setlength{\listparindent}{0in}%

625:  \setlength{\labelsep}{0em}%

626:  \setlength{\labelwidth}{0in}%

627:  \setlength{\itemsep}{12pt}%

628:  \let\makelabel\addendumlabel}

629:  }

630:  {\end{list}\normalsize}

631: 

632: 

633: \newcommand*{\addendumlabel}[1]{\textbf{#1}\hspace{1em}}

634: \begin{addendum}

635:  \item

636: We thank A. Holleitner for making us aware of the details of the

637: experiments being performed with azobenzene, and N. Nemec, D.

638: Tomanek, and R. de Vivie-Riedle for discussions and suggestions.

639: This work was partially funded by the Volkswagen Foundation under

640: grant No.~I/78~340, by the DFG Priority Program ``Quantum Transport

641: at the Molecular Scale'' SPP1243, by the MEC under contracts

642: MAT2005-01388, NAN2004-09109-CO4-04, by the CAM under contract No.

643: S-0505/ESP-0200, and by the European Union project ``Carbon nanotube

644: devices at the quantum limit'' (CARDEQ) under contract No.

645: IST-021285-2. MdV acknowledges the support from the FPI Program of

646: the Comunidad Aut\'{o}noma de Madrid.

647:  \item[Competing Interests] The authors declare that they

648: have no competing financial interests.

649:  \item[Correspondence] Correspondence and requests for

650: materials should be addressed to GC~(email:

651: G.Cuniberti@physik.uni-R.de).

652: 

653: \clearpage

654: \item[Supplementary Information]

655:  To get further intuition on the enhanced

656: conductance in the zigzag-azobenzene junction, we have formulated a

657: minimal single-orbital model (see Fig. S1). The model consists of a

658: semi-infinite armchair (zigzag) CNT, a dimer which mimics the

659: C$_1$-O subsystem at the CNT-molecule interface, and a semi-infinite

660: chain connected to the C$_1$ site, which only fulfills the function

661: of acting as a charge reservoir and will subsequently be treated

662: within the wide-band approximation, \textit{i.e.} assuming an

663: energy-independent coupling between the dimer and the chain.

664: 

665: With this model, we can \textit{qualitatively} describe the local

666: chemistry and topology at the CNT-azobenzene interface. Our main

667: goal is to highlight the influence of the charge state at the

668: ``oxygen'' site --which can be controlled by variations of its

669: onsite energy $\varepsilon_O$-- as well as the role played by the

670: nanotube surface electronic structure.

671: 

672: We can write an analytical expression for the transmission function

673: in terms of the dimer Green function: \beann T(E)= 4\Gamma_R \lcb

674: \Gamma_L^{11}(E) \lsb |G_{C_1C_1}(E)|^2 + |G_{C_1O} (E)|^2 \rsb +\rnothing\\

675: \lnothing 2\Gamma_L^{12}(E) Re(G_{C_1C_1}(E) G_{C_1O}^*(E))\rcb.

676: \eeann

677: 

678: \setcounter{figure}{0}

679: \renewcommand{\figurename}{Figure S}

680: %------------------------------------------------------------------------%

681: \begin{figure}

682: \pippo{\includegraphics[width=.9\linewidth]{Figure_S1.eps}}

683: \myCaption{Transmission spectrum for the $\pi$-orbital model.}{The

684: left panels refer to a (5,5) armchair-C$_1$O dimer-wide band lead

685: (WBL) junction and the left ones to the case in which the nanotube

686: is a (9,0) metallic zigzag tube. The upper panels refer to a full

687: carbon junction with equal onsite energies on the dimer. In the

688: lower panels the onsite energy of the atom indicated as O in the

689: sketches of the junctions has been lowered to mimic the different

690: electronic affinities of carbon and oxygen. The shaded yellow area

691: refers to the integration region of $0.8$ Volt over which we have

692: calculated the $I$-$V$ characteristics in the main text. As it can

693: be clearly seen in the lower panel, the zigzag junction transmission

694: profile does result in a much higher current than for the armchair

695: one inside this voltage window.}

696: \end{figure}

697: %------------------------------------------------------------------------%

698: In this equation, $\Gamma_L^{11}(E)$ and $\Gamma_L^{12}(E)$ describe

699: the direct and cross coupling of the dimer to the left CNT electrode

700: (the indices 1 and 2 refer to the C$_1$ and O atoms, respectively),

701: and $\Gamma_R$ the constant coupling of the C$_1$ atom to the

702: wide-band lead (WBL) electrode on the right. The dimer Green

703: function can be obtained as solution of a Dyson $2x2$ matrix

704: equation $G^{-1}(E) = E-H_{dimer}- \Sigma_L(E)- \Sigma_R(E)$.

705: 

706: Three contributions to the transmission function can be clearly

707: distinguished: $G_{C_1C_1}$ is related to tunnelling pathways going

708: from the CNT directly through the C$_1$ atom, $G_{C_1O}$ describes

709: pathways passing through the O atom and then through the C$_1$ atom,

710: and the last term describes interference effects between both

711: pathways. Our calculations show that $|G_{C_1O}(E)|^2$ is the

712: dominant contribution to $T(E)$ for energies around the Fermi level.

713: Additionally this Green function matrix element is the most strongly

714: affected by a shift of the O onsite energies. Thus, we can

715: approximately write: $T(E\approx E_F) \approx 4 \Gamma_R

716: \Gamma_L^{11} |G_{C_1O}(E)|^2$.

717: 

718: In Fig. S1 we show the influence of a shift in $\varepsilon_O$ on

719: the transmission function for both armchair and zigzag junctions (we

720: take the C$_1$ onsite energy equal to that of the CNT carbon atoms).

721: It turns out that changes in $\varepsilon_O$ lead to a dramatic

722: suppression of the transmission for armchair junctions at low

723: energies (left panel of Fig. S1). For the zigzag junction, on the

724: contrary, electronic states lying slightly above the Fermi energy

725: are shifted down (chemical gating) towards $E_F$ without overtaking

726: it, and thus enter the energy window relevant for the low-bias

727: transport (right panel of Fig. S1). The value  $\varepsilon_O= -4

728: e\textrm{V}$ has been extracted from our DFTB calculation. It is the

729: difference between the $2p$ orbital energies of isolated oxygen and

730: carbon atoms. The effect discussed here is nevertheless stable

731: against variations around this value.

732: 

733: Extended calculations, which are not included here, show that this

734: peculiar behavior of the zigzag junction should be attributed to the

735: well-known surface state at the Fermi energy: with zigzag (ZZ) CNT

736: leads the dimer resonances are ultimately shifted (renormalized) by

737: the real part of the surface self-energy, which can, to a good

738: approximation, be written near the $E_F$ as:

739: $Re\Sigma^{ZZ}(E)\propto(E-E_F)/(\delta^2+(E-E_F)^2)$, inducing

740: \textit{de facto} a pinning of such resonances around the Fermi

741: energy, where the surface state exists with a linewidth $\delta$. In

742: this way the dimer eigenstates shift into a transport sensitive

743: energy region.

744: 

745: \end{addendum}

746: 

747: \end{document}

748: \bye

749: