1: %% LyX 1.3 created this file. For more info, see http://www.lyx.org/.
2: %% Do not edit unless you really know what you are doing.
3: %\documentclass[twocolumn,showpacs,prl,10pt]{revtex4}
4: %\usepackage[english]{babel}
5: %\usepackage{amsmath,amssymb}
6: %\usepackage{times}
7: %\usepackage{epsfig}
8: %latex fon.tex; dvips -o fon.ps fon.dvi; ps2pdf fon.ps fon.pdf
9:
10: \documentclass[twocolumn,english,aps,prb,floatfix]{revtex4}
11: \usepackage[T1]{fontenc}
12: \usepackage[latin1]{inputenc}
13: \usepackage{graphicx}
14: \usepackage{times}
15:
16: \makeatletter
17:
18:
19:
20:
21:
22: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% LyX specific LaTeX commands.
23: %% Bold symbol macro for standard LaTeX users
24: \newcommand{\boldsymbol}[1]{\mbox{\boldmath $#1$}}
25:
26:
27: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% User specified LaTeX commands.
28:
29:
30: \usepackage{bm}
31:
32:
33: \makeatother
34: \begin{document}
35:
36: \title{Conductance of deformable molecules with interaction }
37:
38:
39: \author{J. Mravlje$^{1}$, A. Ram\v{s}ak$^{1,2}$, and T. Rejec$^{1,2,3}$}
40:
41:
42: \affiliation{$^{1}$J. Stefan Institute, Ljubljana, Slovenia}
43:
44:
45: \affiliation{$^{2}$Faculty of Mathematics and Physics, University of Ljubljana,
46: Slovenia}
47:
48:
49: \affiliation{$^{3}$Department of Physics, Ben-Gurion University, Beer-Sheva,
50: Israel}
51:
52: \date{April 18, 2005}
53: \begin{abstract}
54: Zero temperature linear response conductance of molecules with Coulomb
55: interaction and with various types of phonon modes is analysed
56: together with local occupation, local moment, charge fluctuations and
57: fluctuations of molecular deformation. Deformation fluctuations are
58: quantitatively related to charge fluctuations which exhibit similarity
59: also to static charge susceptibility.
60: \end{abstract}
61:
62: \pacs{71.27.+a,73.23.-b,73.23.Hk}
63:
64: \maketitle
65: The evidence for phonon assisted tunneling was found already in early
66: conductance measurements in double-barrier heterostructures.\cite{goldman87}
67: In conductance measurements of nanoscale systems such as quantum dots
68: or real molecules, the Coulomb interaction leads to the Coulomb blockade
69: or the Kondo effects.\cite{park02} The electron-phonon interaction
70: also proved to play an essential role in such systems. In particular,
71: in single organic molecules electronic transport is influenced by
72: vibrational fine structure.\cite{park00,zhitenev02}
73:
74: Phonon degrees of freedom lead to a mass enhancement of a single electron
75: in the empty conduction band. The problem is known as the conventional
76: polaron problem.\cite{holstein59} The local form of the polaron problem
77: arises when the coupling between electrons and phonons is confined
78: to only one site.\cite{hewson74} Here, the main effect of phonons
79: is a narrowing of the level width, analogous to the electron-phonon
80: mass enhancement and is similar to the level width renormalization
81: due to electron-hole pairs.\cite{haldane78} Theoretical investigations
82: of the combined effect, the electron-electron and the electron-phonon
83: interaction, show that quite diverse impurity systems can be described
84: by Anderson model with renormalized effective parameters.\cite{hewson04}
85:
86: Early studies of conductance of various types of quantum systems with
87: electron-phonon interaction were based on the calculation of the transmission
88: probability as a function of incident energy for an electron interacting
89: with Einstein phonons as it tunnels through a double-barrier structure.\cite{wingreen89}
90: The transmission probability for single injected electrons exhibits
91: phonon-assisted transmission resonances -- side-bands -- at energies
92: near the main elastic resonance.\cite{wingreen89,bonca95} Such side-bands
93: appear also in the linear conductance calculation results if the coupling
94: to the Fermi sea in the leads is not correctly taken into account.\cite{zhu03}
95: However, vibrational side-bands would be discernible in non-linear
96: conductance measurements.\cite{bing04,paaske04} Recently, the numerical
97: renormalization group method applied to a single-molecule device\cite{cornaglia04,cornaglia2}
98: showed that the problem can in certain regimes be mapped onto the
99: anisotropic Kondo model.\cite{costi98} Phonon effects in molecular
100: transistors were investigated also in quantal and classical treatment.\cite{mitra04}
101:
102: In this paper we concentrate on the deformation of a molecule in a
103: linear response conductance measurement. In particular, the molecule
104: is attached to the left and right non-interacting lead, schematically
105: presented in Fig.~\ref{cap:Fig1}, and described with the Anderson
106: model
107:
108: \begin{equation}
109: H_{\mathrm{e}}=-\sum_{\left\langle ij\right\rangle \sigma}t_{ij}c_{i\sigma}^{\dagger}c_{j\sigma}+\epsilon_{d}n_{d}+Un_{d\uparrow}n_{d\downarrow},\label{eq:he}\end{equation}
110: where $\left\langle ij\right\rangle $ represents nearest neighbor
111: hopping. In the leads $t_{ij}\equiv t$ and $t_{\pm1,0}=t_{0,\pm1}\equiv t'$
112: is the hopping matrix element from the leads to the molecule. The
113: occupation of the molecule is $n_{d}=\sum_{\sigma}n_{d\sigma}$ with
114: $n_{d\sigma}=d_{\sigma}^{\dagger}d_{\sigma}$, where $d_{\sigma}\equiv c_{0\sigma}$.
115: The deformation of the molecule is modeled with a general form of
116: electron-phonon coupling and the molecule coupled to the leads is
117: described with
118:
119: \begin{eqnarray}
120: H & = & H_{\mathrm{e}}+\sum_{\alpha}\Omega_{\alpha}a_{\alpha}^{\dagger}a_{\alpha}+\sum_{\alpha}\biggl[M_{\alpha}\left(n_{d}-1\right)\nonumber \\
121: & & +N_{\alpha}\sum_{\sigma}\left(d_{\sigma}^{\dagger}c_{-1\sigma}+d_{\sigma}^{\dagger}c_{1\sigma}+\mathrm{h.c.}\right)\biggr]x_{\alpha},\label{eq:heph}\end{eqnarray}
122: where $M_{\alpha}$ and $N_{\alpha}$ are the local and nearest neighbor
123: electron-phonon coupling constants corresponding to arbitrary number
124: of orthogonal vibrational modes with energies $\Omega_{\alpha}$ and
125: corresponding displacements $x_{\alpha}=a_{\alpha}^{\dagger}+a_{\alpha}$.
126:
127: %
128: \begin{figure}
129: \begin{center}\includegraphics[%
130: width=60mm,
131: keepaspectratio]{Fig1.eps}\end{center}
132:
133:
134: \caption{\label{cap:Fig1}Non-interacting leads attached to an Anderson site,
135: e.g., molecule (gray-shaded site) with various phonon modes.}
136: \end{figure}
137:
138:
139: The zero temperature linear response conductance through the molecule
140: is calculated from the sine formula,\cite{rr03,rrfon} $G=G_{0}\sin^{2}[(E_{+}-E_{-})/4tL]$,
141: where $G_{0}=2e^{2}/h$ and $E_{\pm}$ are the ground state energies
142: of a large auxiliary ring consisted of $L$ non-interacting sites
143: and an embedded interacting system (molecule), with periodic and antiperiodic
144: boundary conditions, respectively. The chemical potential is set at
145: the middle of the band in the leads, which corresponds to $L$ electrons
146: in the system. To determine the ground state energy, we generalized
147: the projection-operator method proposed by Gunnarsson and Sch\"{ o}nhammer.\cite{gunnarsson85}
148: The variational expression for the ground state is
149:
150: \begin{equation}
151: |\Psi\rangle=\sum_{\lambda\left\{ \! m_{\alpha}\!\right\} }C_{\lambda\left\{ \! m_{\alpha}\!\right\} }P_{\lambda}\prod_{\alpha}a_{\alpha}^{\dagger m_{\alpha}}\left|\tilde{0}\right\rangle ,\label{eq:psi}\end{equation}
152: where $P_{\lambda}$ are projection operators to multi-electron states
153: of an isolated molecule, $P_{0}=\left(1-n_{d\uparrow}\right)\left(1-n_{d\downarrow}\right)$,
154: $P_{1}=\sum_{\sigma}n_{d\sigma}\left(1-n_{d\bar{\sigma}}\right)$,
155: and $P_{2}=n_{d\uparrow}n_{d\downarrow}$, as well as additional operators
156: involving hopping of electrons between the molecule and leads (for
157: details, see Ref.~\onlinecite{gunnarsson85}). The vacuum state $\left|\tilde{0}\right\rangle $
158: is the ground-state of a decoupled, non-interacting electron-phonon
159: system, described by renormalized matrix elements $\tilde{t}^{\prime}$
160: and $\tilde{\epsilon}_{d}$. An approximation to the ground-state
161: energy is obtained by minimizing the total energy with respect to
162: coefficients $C_{\lambda\left\{ \! m_{\alpha}\!\right\} }$ and parameters
163: $\tilde{t}^{\prime}$ and $\tilde{\epsilon}_{d}$ while allowing a
164: sufficiently large number of excited phonons, in order to obtain a
165: converged result.
166:
167: In the limit of large frequencies, $\Omega_{\alpha}$, finite $M_{\alpha}$
168: and $N_{\alpha}=0$, the model is equivalent to the bare Anderson
169: model with renormalized parameters $U_{\mathrm{eff}}=U-\sum_{\alpha}2M_{\alpha}^{2}/\Omega_{\alpha}$
170: and $\epsilon_{d,\mathrm{eff}}=\epsilon_{d}-\sum_{\alpha}M_{\alpha}^{2}/\Omega_{\alpha}$.\cite{hewson74}
171: We first test our numerical formalism in this regime. In Fig.~\ref{cap:Fig2}(a)
172: the conductance through an undeformable molecule ($M_{\alpha}=N_{\alpha}=0$),
173: for various $U$ and a fixed $\Delta=2t^{\prime2}/t=t/5$ is presented.
174:
175: %
176: \begin{figure}
177: \begin{center}\includegraphics[%
178: width=75mm]{Fig2.eps}\end{center}
179:
180:
181: \caption{\label{cap:Fig2} (Color online) (a) Conductance for the bare Anderson
182: model with $-6\Delta\leq U\leq10\Delta$ in increments of $2\Delta$
183: (full lines for $U>0$, dashed lines for $U<0$ and a thicker full
184: line for $U=0$). (b) Local occupancy $n$ and local moment $M_{\mathrm{{loc}}}$
185: (inset). (c) Charge fluctuations $\Delta n^{2}=2n-n^{2}-M_{\mathrm{{loc}}}^{2}$.
186: Inset: renormalized charge susceptibility $(\pi\Delta/4)\chi_{c}$.}
187: \end{figure}
188:
189:
190: For a positive $U$ the conductance exhibits a plateau in the Kondo
191: regime and the results obtained with the present method accurately
192: reproduce\cite{rr03} the exact solution based on the Bethe ansatz.\cite{wiegman80}
193: Local electron density $n=\left\langle n_{d}\right\rangle $ is related
194: to the conductance through the Friedel sum rule\cite{friedel} and
195: is characterized with a plateau $n\sim1$ in the Kondo regime, as
196: presented in Fig.~\ref{cap:Fig2}(b). Kondo physics is signalled
197: also with the increase of the local moment $M_{\mathrm{{loc}}}=\bigl<\left(n_{d\uparrow}-n_{d\downarrow}\right)^{2}\bigr>^{1/2}$,
198: presented in the inset. The related occupancy (charge) fluctuations,
199: $\Delta n^{2}=\bigl<\left(n_{d}-n\right)^{2}\bigr>$, are presented
200: in Fig.~\ref{cap:Fig2}(c) and in the inset is given the corresponding
201: charge susceptibility, $\chi_{c}=-\partial n/\partial\epsilon_{d}$.
202: In accordance with the fluctuation-dissipation theorem, charge fluctuations
203: are similar to the charge susceptibility, $\Delta n^{2}\sim(\pi\Delta/4)\chi_{c}$.\cite{fdt}
204: Fluctuations are suppressed in the Kondo regime, and are larger in
205: the mixed valence regime, $\left|\epsilon_{d}\right|\lesssim\Delta$
206: or $\left|\epsilon_{d}+U\right|\lesssim\Delta$.
207:
208: For sufficiently strong electron-phonon coupling $M_{\alpha}$ the
209: effective electron-electron interaction is attractive, $U_{\mathrm{eff}}<0$.
210: In this regime the situation is opposite to the more standard spin
211: Kondo regime because at the impurity the system favors electron (hole)
212: pairs rather than local moments due to single electrons.\cite{sch84}
213: Therefore strong charge fluctuations emerge in the particle hole symmetric
214: point when the chemical potential is level with the local bipolaron
215: energy leading to charge-fluctuation (anisotropic) Kondo effect.\cite{costi98}
216: In Fig.~\ref{cap:Fig2}(a) the conductance for various attractive
217: $U<0$ in the bare Anderson model is presented with dashed lines.
218: The first observation is a narrowing of the conductance curve and
219: the corresponding enhanced charge fluctuations {[}Fig.~\ref{cap:Fig2}(c){]},
220: consistent with a sharp transition in the local occupation and a suppresion
221: of the local moment, Fig.~\ref{cap:Fig2}(b). For increasing (negative)
222: $U$, charge susceptibility is not limited and overshoots charge fluctuations.
223: However, the comparison of $\Delta n^{2}$ and $\chi_{c}$ confirms
224: at least qualitative proportionality. Our analysis showed that these
225: results for the renormalized bare Anderson model represent also generic
226: behavior of the general model with $N_{\alpha}=0$.
227:
228: In Fig.~\ref{cap:Fig3} the results of the analysis of a molecule
229: with a single vibrational mode using $U=10\Delta$ are presented.
230: The coupling--frequency ratio is kept constant, $M_{\alpha}/\Omega_{\alpha}\equiv M/\Omega=1$,
231: while the electron-phonon coupling $M$ is varied from $M=\Delta$
232: to $M=6\Delta$ in increments of $\Delta$. The results for conductance,
233: occupancy and occupancy fluctuations {[}Figs.~\ref{cap:Fig3}(a,
234: b, c){]} resemble the results of the bare Anderson model with renormalized
235: parameters.\cite{hewson04} There are no discernible side-bands in
236: the conductance. For comparison also the result for undeformable molecule
237: $(M=0)$ is presented. The molecule deformations, i.e. $x_{\alpha}$,
238: are related to the occupation of the impurity. The displacement of
239: individual modes is in general given with $\langle x_{\alpha}\rangle=
240: -2M_{\alpha}/\Omega_{\alpha}(n-1)-16N_{\alpha}/\Omega_{\alpha}\textrm{Re}
241: \langle d_{\sigma}^{\dagger}c_{1\sigma}\rangle$.\cite{cornaglia2}
242: Fluctuations of the deformation, $\Delta x_{\alpha}^{2}=\bigl<\left(x_{\alpha}-\left\langle x_{\alpha}\right\rangle \right)^{2}\bigr>$,
243: are related to the average number of particular phonons in the system,
244: $\Delta x_{\alpha}^{2}=1+2\left\langle a_{\alpha}^{\dagger}a_{\alpha}\right\rangle +2\textrm{Re}\left\langle a_{\alpha}^{\dagger}a_{\alpha}^{\dagger}\right\rangle -\left\langle x_{\alpha}\right\rangle ^{2}$.
245: A deviation of $\Delta x_{\alpha}^{2}$ from unity signals deviations
246: from the coherent state of the oscillator. In the limit of large phonon
247: frequencies (fast modes) the oscillator deformations can follow charge
248: fluctuations, $\Delta x_{\alpha}^{2}-1=(2M_{\alpha}/\Omega_{\alpha})^{2}\Delta n^{2}$,
249: while in general they are smaller than that. In the limit of small
250: phonon frequencies (slow modes), phonons feel the average occupation
251: of the molecule and $\Delta x_{\alpha}^{2}-1=0$. The fluctuations
252: corresponding to a single mode system, $\Delta x_{\alpha}^{2}\equiv\Delta x^{2}$,
253: for the same set of parameters and labeled as above, are presented
254: in Fig.~\ref{cap:Fig3}(d).
255:
256: %
257: \begin{figure}
258: \begin{center}\includegraphics[%
259: width=90mm]{Fig3.eps}\end{center}
260:
261:
262: \caption{\label{cap:Fig3} (Color online) Results for $U=10\Delta$ and a
263: single phonon mode with a fixed $M=\Omega$. (a) Conductance vs. $\left(\epsilon_{d}+U/2\right)/\Delta$
264: for $0\leq M\leq6\Delta$ in increments of $\Delta$. The dotted line
265: represents the results for an undeformable molecule, $M=0$. (b) Occupation,
266: (c) occupation fluctuations and (d) deformation fluctuations. }
267: \end{figure}
268:
269:
270: Next we turn our attention to the case of a molecule with multiple
271: vibrational modes. Here we give results for the case $U=10\Delta$
272: with mode 1 with frequency $\Omega_{1}=\Delta$ and mode 2 with frequency
273: $\Omega_{2}=10\Delta$. The effective Coulomb interaction is reduced
274: due to coupling to both modes $U_{\mathrm{eff}}=U-2M_{1}^{2}/\Omega_{1}-2M_{2}^{2}/\Omega_{2}$.
275: We take $N_{\alpha}=0$ and thus the system retains the particle-hole
276: symmetry. Therefore, in Fig.~\ref{cap:Fig4} only $\epsilon_{d}+U/2>0$
277: regime is shown. In order to study both, particular and mutual influence
278: of different modes, we fix $U_{\mathrm{eff}}\equiv5\Delta$, and set
279: $2M_{1}^{2}/\Omega_{1}=r\left(U-U_{\mathrm{eff}}\right)$, $2M_{2}^{2}/\Omega_{2}=\left(1-r\right)\left(U-U_{\mathrm{eff}}\right)$
280: where by varying $r$ one can explore the effect of particular modes.
281: For a single stiffer mode 2, $r=0$, the conductance curve, Fig.~\ref{cap:Fig4}(a),
282: is suppressed in the Kondo regime and enhanced in the empty orbital
283: region compared to the softer mode 1, $r=1$. As a limiting case of
284: this regime the bare Anderson model results for $U=5\Delta$ are presented.
285:
286: As a representative of the opposite limit, we consider very soft phonons
287: with $\Omega=\Delta/100$. In the Kondo regime the conductance is
288: close to the unrenormalized Anderson model result with $U=10\Delta$.
289: In the mixed valence regime the curve is much steeper, due to strong
290: renormalization of hopping parameter $\tilde{t}^{\prime}$. In the
291: empty-orbital regime the conductance approaches the result obtained
292: with doubly reduced electron-electron interaction $\tilde{U}=U-4M^{2}/\Omega$,
293: which can be understood as follows. First the oscillator displacement
294: is shifted, $x\rightarrow\tilde{x}+2\lambda$ thus the Hamiltonian
295: is transformed into $\tilde{H}=\left(\epsilon_{d}+2\lambda M\right)n_{d}+\tilde{x}\left[M\left(n_{d}-1\right)+\Omega\lambda\right]+\Omega\tilde{a}^{\dagger}\tilde{a}+...$,
296: where $\lambda=-M\left(n-1\right)/\Omega$, with vanishing transformed
297: displacement. This Hamiltonian can be solved with trial wave functions
298: with $m_{\alpha}=0$. Renormalized local energies are then $\epsilon_{d}+2M^{2}/\Omega$,
299: $\epsilon_{d}$, and $\epsilon_{d}-2M^{2}/\Omega$ for $n=0,1,2$,
300: respectively. The shifts of $\epsilon_{d}$ where $n=0,\,2$ in turn
301: correspond to reduced $\tilde{U}=U-4M^{2}/\Omega$ and to $\tilde{U}=U$
302: for $n=1$.
303:
304: %
305: \begin{figure}
306: \begin{center}\includegraphics[%
307: width=70mm]{Fig4.eps}\end{center}
308:
309:
310: \caption{\label{cap:Fig4} (Color online) A fixed $U_{\mathrm{eff}}=5\Delta$
311: with $U=10\Delta$ and $\Omega_{1}=\Delta$, $\Omega_{2}=10\Delta$
312: for various $M_{1,2}$ (corresponding to $r=0$, 1/4, 1/2, 3/4 and
313: 1) - full lines. Also plotted are the results for a bare Anderson
314: model with $U=10\Delta$, $U=5\Delta$ and $U=0$ (dotted, short-dashed
315: and dashed-dotted, respectively). Long-dashed lines correspond to
316: a single softer mode with $\Omega=\Delta/100$ and the same $U_{\mathrm{eff}}=5\Delta$.
317: (a) Conductance, (b) occupation fluctuations and (c) deformation fluctuations
318: for modes 1 and 2. In the inset, the deformation fluctuations for
319: a single softer mode are shown.}
320: \end{figure}
321:
322:
323: Occupancy $n$ is related to the conductance similarly as in the previous
324: figures and is not presented for this case. Charge fluctuations, Fig.~\ref{cap:Fig4}(b),
325: are similar to the relation between $\Delta n^{2}$ and $\chi_{c}$
326: as in the above single mode case: the fluctuations are larger for
327: stiffer phonon modes, except in the mixed-valence regime, where $\Delta n^{2}$
328: is very weakly dependent of $\Omega_{\alpha}$. Charge fluctuations
329: for the case of softer mode $\Omega=\Delta/100$ exhibit limiting
330: behavior consistent with $G$ as discussed above. Deformation fluctuations,
331: $\Delta x_{\alpha=1,2}^{2}$, are shown in Fig.~\ref{cap:Fig4}(c).
332: As expected, the fluctuations of the softer mode 1 are enhanced in
333: comparison with the stiffer mode 2. This effect is even more pronounced
334: for $\Omega=\Delta/100$ {[}inset of Fig.~\ref{cap:Fig4}(d){]}.
335:
336: %
337: \begin{figure}
338: \begin{center}\includegraphics[%
339: width=65mm]{Fig5.eps}\end{center}
340:
341:
342: \caption{\label{cap:Fig5} (Color online) (a) Conductance, (b) occupation
343: fluctuations, and (c) deformation fluctuations for various types of
344: electron-phonon interaction ($U=10\Delta$, $\Omega=5\Delta$).}
345: \end{figure}
346:
347:
348: In Fig.~\ref{cap:Fig5} we present the results for a general case
349: of single electron-phonon mode coupling with $U=10\Delta$ and $\Omega=5\Delta$.
350: As pointed out by Cornaglia \emph{et al.}\cite{cornaglia2}, $N_{\alpha}$
351: terms together with $M_{\alpha}$ break the particle-hole symmetry,
352: while the symmetry is conserved if only one of the terms is non-vanishing.
353: In Fig.~\ref{cap:Fig5}(a) the conductance for three typical cases
354: is shown. Full line represents the $M=0$ and $N=5\Delta$ results
355: where the resonance width is severely increased, because the $N$-terms
356: increase the effective $t'$. If both electron-phonon coupling terms
357: are relevant, e.g., $M=3.5\Delta$ and $N=3\Delta$, the conductance
358: exhibits asymmetry (dashed line), compared to the $N=0$ case (dotted
359: line). In Fig.~\ref{cap:Fig5}(b) the corresponding occupation fluctuations
360: are presented. Displacement fluctuations $\Delta x_{\alpha}$, Fig.~\ref{cap:Fig5}(c),
361: in this case are not related solely to occupation fluctuations, but
362: also to fluctuations of the hopping operator $d_{\sigma}^{\dagger}c_{1\sigma}$
363: (not shown here).
364:
365: We have presented results of a comprehensive analysis of linear response
366: conductance through a deformable molecule with electron-electron interaction
367: and different orthogonal phonon modes. In general, the conductance
368: does not exhibit side-bands and is related to the Anderson model with
369: renormalized parameters for the single- or multiple-phonon interaction.
370: Additionally, the emphasis of our analysis was on the deformation
371: fluctuations of the molecule due to multiple phonon modes. Phonons
372: in slow phonon modes are permanently in a coherent state corresponding
373: to the average occupation of the molecule and the deformation fluctuations
374: are minimal in this limit (except close to the charge fluctuations
375: maxima). In the opposite limit of a fast phonon mode, phonons form
376: a coherent state corresponding to the occupation at a given moment
377: in time. Therefore, deformation fluctuations are enhanced proportionally
378: to charge fluctuations, the proportionality coefficient being $-(2M_{\alpha}/\Omega_{\alpha})^{2}$.
379: In general, the deformation fluctuations take a value between these
380: two limits. It was also shown that charge fluctuations are approximately
381: proportional to static charge susceptibility of the molecule. The
382: method used here proved to be robust and appropriate for a wide range
383: of generalizations due to specific electron-phonon interaction or
384: the topology of the interaction region, for example, a molecule with
385: several interconnected sites.
386:
387: One of the authors (A.R.) would like to thank A. Hewson for stimulating
388: discussions. The authors acknowledge the support of the MSZS under
389: grant Pl-0044.
390:
391: \begin{thebibliography}{10}
392: \bibitem{goldman87}V.~J.~Goldman, D.~C.~Tsui, and J.~E.~Cunningham, Phys. Rev.
393: B \textbf{36}, 7635 (1987).
394: \bibitem{park02}J.~Park \emph{et al.}, Nature \textbf{417}, 722 (2002).
395: \bibitem{park00}J.~Park \emph{et al}., Nature \textbf{407}, 57 (2000).
396: \bibitem{zhitenev02}N.~B.~Zhitenev, H.~Meng, and Z.~Bao, Phys. Rev. Lett. \textbf{88},
397: 226801 (2002).
398: \bibitem{holstein59}T.~Holstein, Ann. Phys. NY \textbf{8}, 343 (1959).
399: \bibitem{hewson74}A.~C. Hewson and D.~M.~News, J. Phys. C \textbf{13}, 4477 (1980).
400: A.~C.~Hewson and D.~Meyer, J. Phys.: Condens. Matter \textbf{14},
401: 427 (2002).
402: \bibitem{haldane78}F.~D.~M.~Haldane, \emph{Thesis}, Cambridge University (1978).
403: \bibitem{hewson04}A.~C.~Hewson, A.~Oguri, and D. Meyer, Eur. Phys. J. B \textbf{40},
404: 177 (2004).
405: \bibitem{wingreen89}N.~S.~Wingreen, K.~W.~Jacobsen, and J.~W.~Wilkins, Phys. Rev.
406: B \textbf{40}, 11834 (1989).
407: \bibitem{bonca95}J.~Bon\v{c}a and S.~A.~Trugman, Phys. Rev. Lett \textbf{75}, 2566
408: (1995).
409: \bibitem{zhu03}Jian-Xin Zhu and A.~V.~Balatsky, Phys. Rev. B \textbf{67}, 165326
410: (2003).
411: \bibitem{bing04}Bing Dong, H.~L.~Cui, and X.~L.~Lei, Phys. Rev. B \textbf{69},
412: 205315-1 (2004).
413: \bibitem{paaske04}J. Paaske and K. Flensberg, cond-mat/0409158; S.~Braig and K.~Flensberg,
414: Phys. Rev. B \textbf{68}, 205324 (2003).
415: \bibitem{cornaglia04}P.~S.~Cornaglia, H.~Ness, D.~R.~Grempel, Phys. Rev. Lett. \textbf{93},
416: 147201 ( 2004).
417: \bibitem{cornaglia2}P.~S.~Cornaglia, D.~R.~Grempel, and H.~Ness, cond-mat/0409021.
418: \bibitem{costi98}T.~A.~Costi, Phys. Rev. Lett. \textbf{80}, 1038 (1998).
419: \bibitem{mitra04}A.~Mitra, I.~ Aleiner, A.~J.~Millis, Phys. Rev. B \textbf{69},
420: 245302 (2004).
421: \bibitem{rr03}T.~Rejec and A.~Ram\v{s}ak, Phys. Rev. B \textbf{68}, 033306 (2003).
422: \bibitem{rrfon}In Ref.~\onlinecite{rr03} a general proof that the method is correct
423: in the case of the electron-electron interaction is given. The main
424: assumption there is the existence of the Fermi liquid properties of
425: the interacting system coupled to the leads. For the case of the electron-phonon
426: coupling studied here the main steps of the proof can be repeated
427: and the conclussions remain unchanged.
428: \bibitem{gunnarsson85}O.~Gunnarsson and K.~Sch\"{ o}nhammer, Phys. Rev. B \textbf{31},
429: 4815 (1985).
430: \bibitem{wiegman80}P.~B.~Wiegman, Sov. Phys. JETP Lett. \textbf{31}, 392 (1980); N.
431: Andrei, Phys. Rev. Lett. \textbf{45}, 379 (1980).
432: \bibitem{fdt}Strictly, $\langle n_{d}^{2}\rangle$ is given with the integral of
433: imaginary part of dynamic charge susceptibility, $\chi_{c}^{''}(\omega)$,
434: therefore the relation to static $\chi_{c}$ is only qualitative (even
435: at $U=0$). We should note that for large $U$ limit we found reasonable
436: agreement with $1/{\cal N}$-expansion result $\chi_{c}\sim\frac{\pi}{{\cal N}\Delta}n^{2}(1-n).$
437: \bibitem{friedel}$G=G_{0}\sin^{2}\frac{\pi}{2}n$ for $t\to\infty$.
438: \bibitem{sch84}K.~Sch\"{ o}nhammer and O.~Gunnarsson, Phys. Rev. B
439: \textbf{30}, 3141 (1984).
440: \end{thebibliography}
441:
442: \end{document}
443: