cond-mat0609021/main.tex
1: %\documentclass[showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
2: \documentclass[showpacs,superscriptaddress,twocolumn,aps,amsmath,amssymb]{revtex4}
3: 
4: \usepackage{amsfonts}
5: \usepackage{amsmath}
6: \usepackage{graphicx}
7: \usepackage{dcolumn}
8: \usepackage{bm}
9: 
10: \newcommand{\be}{\begin{equation}}
11: \newcommand{\ee}{\end{equation}}
12: \newcommand{\bea}{\begin{eqnarray}}
13: \newcommand{\eea}{\end{eqnarray}}
14: \newcommand{\Fig}[1]{Fig.\ref{#1}}
15: \newcommand{\Eq}[1]{Eq.\,(\ref{#1})}
16: \newcommand{\Eqs}[1]{Eqs.\,(\ref{#1})}
17: \newcommand{\dg}{\dagger}
18: \newcommand{\la}{\langle}
19: \newcommand{\ra}{\rangle}
20: \newcommand{\nl}{\nonumber \\}
21: \newcommand{\eph}{\emph{e-ph}\ }
22: \newcommand{\Tab}[1]{Table\ \ref{#1}}
23: \newcommand{\Sec}[1]{Sec.\,\ref{#1}}
24: \newcommand{\Sch}{Schr\"{o}dinger\ }
25: \newcommand{\mbpar}[1]{\left( #1 \right)}
26: \newcommand{\mint}[2]{\int\limits_{#1}^{#2}}
27: 
28: %\bibliographystyle{acs}
29: %\bibliographystyle{apsrev}
30: %\bibliographystyle{pf}
31: 
32: \begin{document}
33: 
34: 
35: \title{Bias-induced insulator-metal transition in organic electronics}
36: \author{J.~H.~Wei}
37: \affiliation{Department of Physics, Renmin University of China,
38: Beijing, China} \affiliation{Department of Physics, Shandong
39: University, Jinan, China}
40: 
41: \author{S.~J.~Xie}
42: \affiliation{Department of Physics, Shandong University, Jinan,
43: China}
44: 
45: \author{L.~M.~Mei}
46: \affiliation{Department of Physics, Shandong University, Jinan,
47: China}
48: 
49: \author{YiJing Yan}
50: \affiliation{Department of Chemistry, Hong Kong University
51:    of Science and Technology, Kowloon, Hong Kong}
52: 
53: \date{\today}
54: 
55: \begin{abstract}
56:  We investigate the bias-induced insulator-metal transition
57:  in organic electronics devices, on the basis of the
58:  Su-Schrieffer-Heeger model [W.~P. Su, J.~R. Schrieffer, and A.~J. Heeger,
59:  Phys. Rev. B {\bf 22}, 2099 (1980)] combined with the non-equilibrium
60:  Green's function formalism. The insulator-metal
61:  transition is explained with the energy levels
62:   crossover that eliminates the Peierls phase [R.~Peierls,
63: {\em Quantum Theory of Solids}, Oxford University Press, Oxford,
64: 1955] and
65:  delocalizes the electron states near the
66:  threshold voltage. This may account for
67:  the experimental observations on the devices
68:  that exhibit intrinsic bistable conductance switching
69:  with large on-off ratio.
70: \end{abstract}
71: 
72: \pacs{71.30.+h, 72.80.Le, 73.63.-b}
73: 
74: \maketitle
75: %
76: \clearpage
77: 
78:   In recent years molecular electronics,
79:  which uses individual molecules as possible electrical switches,
80:  has grown up rapidly since the discoveries of negative
81:  differential resistance and
82:  bistable conductance switching (CS) \cite{Che991550,Don012303}.
83:  Individual molecules can be wired to serve as diodes, transistors,
84:  and other electronic devices at the heart of
85:  computer chips \cite{Ree013735,Col001172,Kra022927,Raw03377}.
86:  For the next generation of technology,
87:  molecular electronics should exhibit
88:  a large on-off ratio ($>$\,50:1).
89:  It should also be intrinsic and controllable, rather than
90:  artificial due to for example metal filament \cite{Ser03556}.
91:   Some mechanisms such as the conformational change \cite{Don012303},
92:  charging effect and polaron models \cite{Gal05125,Wei0682} have
93:  been proposed. However, none of them can fully account
94:  for the large intrinsic on-off ratio
95:  of molecular switching observed
96:  in experiments \cite{Blu05167,Kum991645,Mat0512450}.
97:  In this work, we shall propose another possible
98:  mechanism, in which the
99:  intrinsic CS with large on-off ratio
100:  is controlled by the bias-triggered
101:  Peierls phase elimination and energy levels crossover.
102: 
103:   Let us start with the following observations.
104:  Most molecular devices use short-chain $\pi$-conjugated
105:  organic molecules as the active region, and are
106:  fabricated as metal-molecule-metal sandwiched structures.
107:  A chain-like nanoscale organic molecule belongs to
108:  the family of organic semiconductors (OSEs).
109:  Physically, an OSE is charactered by the Peierls instability;
110:  i.e., the strong electron-phonon ({\it e-ph})
111:  interaction that induces the single--double bond alternation
112:  and a relative large energy gap between highest-occupied and
113:  lowest-unoccupied molecular orbital (HOMO and LUMO) \cite{Pei55}.
114:  It is the Peierls instability that largely localizes
115:  the $\pi$-electrons of carbon and contributes to
116:  the low conductance of OSEs, no matter
117:  they are fabricated via the bottom-up self-assembled monolayer
118:  or the top-down lithography technique.
119:  Apparently, delocalizing $\pi$-electrons will improve
120:  the conductivity of OSEs. One method is the creation of
121:  nonlinear excitations (solitons or polarons) to act as charge
122:  carriers \cite{Su802099}.  Another approach
123:  is to eliminate the Peierls phase directly.
124:   Increasing temperature may
125:   melt the Peierls phase, leading to an insulator--metal
126:  (I-M) transition \cite{Pei55}.
127: %
128: 
129:  An effective and controllable method
130:  in electronics should however be the one triggered
131:  by the bias or other external control field.
132: %%%
133:  In this connection and also in contact
134:  with some typical
135:  experimental systems \cite{Kum991645,Mat0512450},
136:  we present in this work a theoretical study
137:  on the bias-induced I-M transition in a model
138:  nanoscale metal/OSE/metal sandwich structure.
139: %
140:  We will elucidate the fact that the bias-triggered
141:  energy levels crossover and its resulting Peierls phase
142:  elimination can largely account for the observed CS.
143: 
144:  We combine the celebrated Su-Schrieffer-Heeger (SSH) model \cite{Su802099}
145:  and the nonequilibrium Green's function (NEGF)
146:  formalism \cite{Kel641515,Bra02165401} to study organic
147:  electronics. For the spin-independent charge transport considered
148:  in this work, the SSH Hamiltonian
149:  for the OSE electrons coupled adiabatically with the lattice
150:  displacements assumes
151: \bea\label{SSH}
152:  H_{O}&=&\sum_{n}
153:   \Bigl\{\epsilon _{o}c_{n}^{+}c_{n}-
154:   [t_{o}-(-1)^{n}t_{1}-\alpha _{o}y_{n}]\times
155: \nl &\ &
156:   (c_{n}^{+}c_{n+1}+
157:   c_{n+1}^{+}c_{n})\Bigr\}+\frac{K_{o}}{2}\sum_{n}y_{n}^{2}.
158: \eea%
159:  Here, $c_{n}^{+}$ ($c_{n}$) denotes the creation
160: (annihilation) operator of an electron at the $n^{\mathrm{th}}$
161:  site of the OSE, and  $\epsilon _{o}$ the on-site energy,
162:  $t_{o}$ the zero-displacement hopping integral,
163:  and $t_{1}$ the nondegeneracy parameter,
164:  respectively. The lattice distortion is written
165:  in terms of the bond distances $\{y_{n}=u_{n+1}-u_{n}\}$
166:  and treated classically, with
167:  the spring constant $K_{o}$
168:  and the adiabatic {\it e-ph}\
169:  coupling constant $\alpha _{o}$.
170:  The L and R electrodes are chosen to be an identical nonmagnetic metal.
171:  They are individually described
172:  by the one-dimensional single-band tight-binding Hamiltonian,
173:  being of the on-site energy $\epsilon _{f}$ and the
174:  nearest neighbor transfer integral $t_{f}$.
175: 
176:  At the mean-field level of correlations, one can evaluate the
177:  current based on the NEGF formalism \cite{Kel641515,Bra02165401},
178: \begin{equation} \label{I-V}
179:   I=\frac{2e}{h}\int\limits_{-\infty }^{\infty }
180:   {\mbox{Tr}\left( \Gamma _{\mathrm{L}}G^{\rm r}
181:     \Gamma _{\mathrm{R}}G^{\rm a}\right) }
182:    [f(E,\mu _{\mathrm{L}})-f(E,\mu _{\mathrm{R}})]dE.
183: \end{equation}%
184:  Here, $f(E,\mu _{\mathrm{L/R}})$ is the Fermi distribution
185:  function at the lead chemical potential $\mu _{\mathrm{L/R}}$.
186:  The trace term in the integrand is the transmission coefficient
187:  function $T(E)$. The involving
188:  $\Gamma _{\mathrm{L/R}}$ denotes the lead reservoir-induced
189:   broadening matrix, while
190:   $G^{\rm r}$ and  $G^{\rm a}=(G^{\rm r})^{\dg}$ are the retarded
191:  and the advanced single-particle Green's functions
192:  for the central scattering region (S-region),
193:  under a finite applied bias. This region
194:  consists of the OSE together with a number of
195:  metal atoms attached to each of its ends.
196: %%%
197:    For a quantum open system, the effects
198:  of both the charging from the electrodes and the
199:  external potential from the bias voltage
200:  are contained in the
201:  nonequilibrium reduced density matrix,
202:  $\rho =\frac{1}{2\pi }\sum_{\alpha
203:   =\mathrm{L,R}}  %\int\limits_{-\infty }^{\infty }
204:    \int G^{\rm r}\Gamma _{\alpha}G^{\rm a}
205:    f(E,\mu _{\alpha})dE$,
206:  of the coupled {\it e-ph}\ system \cite{Bra02165401}.
207:  The reduced density matrix should be evaluated
208:  in a self-consistent manner,
209:   together with the Poisson's
210:  equation and the Hellman-Feynman (H-F)
211:  variation theorem for lattice distortion:
212:  $\partial \lbrack \mathrm{Tr}%
213:  (H_{\mathrm{S}}\rho )]/\partial u_{n}=0$. The self-consistent
214:  procedure as well as the time-to-voltage mapping method to handle
215:  the time-dependent bias sweeping have been described
216:  in detail elsewhere \cite{Wei0682,Wei0508417}.
217: 
218:    In the following calculation of the hysteretic $I-V$ characteristics,
219: we choose a constant rate of bias sweeping to be $0.1$ V/sec, as
220: both the numerical efficiency and the typical experimental value
221: being of $0.01\sim 0.1$ V/sec are concerned. Note also that
222: SSH+NEGF formulation includes two main approximations. The first
223: one is the static lattice description involving in the model
224: Hamiltonian and the aforementioned H-F variation theorem. This
225: approximation is valid when the lattice fluctuation time scale is
226: longer than the inverse coherent charge transfer coupling. The
227: second approximation is the meanfield treatment of
228: electron-electron correlation that involves in the NEGF
229: formulation and also the model Hamiltonian. This is a common
230: approach, especially in treating strong
231:  {\it e-ph} systems.
232: 
233:  We choose the SSH parameters in \Eq{SSH} the
234: typical values:
235:  $t_{o}=2.5$ eV, $t_{1}=0.04$ eV, $\alpha _{o}=4.2$ eV/\r{A},
236:  and $K_{o}=21.0$ eV/\r{A}$^{2}$.
237:  The transfer integral for metal electrodes
238:  is $t_{f}=3.0$ eV to produce a wide
239:  band (4$t_{f}$), and the metal-OSE
240:  coupling transfer integral is
241:  $t_{\rm metal-OSE}=0.5(t_{o}+t_{f})$.
242:  Without losing generality,
243:  we set  $E_{\rm F}=\epsilon _{o}=\epsilon_{f}=0$,
244:  where $E_{\rm F}$ is the equilibrium Fermi level at
245:  the zero bias $V=0$.
246: 
247: %\input{figcap1}
248: 
249:  We then calculate the nonlinear current [\Eq{I-V}] as a function
250:  of sweeping bias of the model metal/OSE/metal electronics at
251:  $T=11$ K by using the time-to-voltage mapping method \cite{Wei0508417}.
252:  The resulting $I$-$V$ curve, presented in
253: \Fig{fig1}, clearly shows hysteretic CS characters.
254:  Note that the model OSE is of $N_o=40$, which is about 5nm
255:  and comparable to the
256:  experiments in Refs.\ \onlinecite{Blu05167} and
257:  \onlinecite{He0663}.
258:  More than 30 years ago, Simmons
259:  \emph{et al.}\ reported an
260:  electroformed metal-insulator-metal diode
261:  with reversible voltage memory effects \cite{Sim6777}.
262:  Recently, Bozano {\it et al.}\
263:  had also demonstrated a similar resistive
264:  switching phenomenon in OSE layers \cite{Boz04607}.
265:  They had proposed several necessary features of an operational
266:  switch, as summarized in Ref.\ \onlinecite{Boz04607}.
267:  In comparison with their experiments,
268:  especially the Fig.\,3
269:  in Ref.\ \onlinecite{Boz04607},
270:  our model calculation dose possess most of the necessary features
271:  they proposed. These include
272:  (1) a distinct bistable on/off state
273:  corresponding to low/high resistance with
274:   an on-off ratio $\gtrsim$ 20:1;
275:  (2) a local maximum in current $I_{\rm max}$ which
276:  nonlinearly depends on the size of OSE and the metal-OSE
277:  coupling (varying from several nA $\sim{\rm \mu}$A) at voltage $V_{\rm
278:  max}$ ($\sim$ 4.5 V);
279:  (3) a threshold of voltage $V_{\rm th}$
280:  ($\sim$ 3.7 V) at which the high-current
281:  (switch-on) state is established;
282:  (4) reproducible switching and reading which can be
283:  achieved with applied positive or negative voltage.
284: %%
285:   The insert in \Fig{fig1} shows the
286:   nonlinear dependence of the threshold voltage
287:   $V_{\rm th}$ on the OSE chain length $N_o$.
288:  This confirms the generality of our results
289:  in organic electronics devices with different sizes.
290: 
291:  Let us discuss the mechanism of the bistable CS,
292:  as indicated in \Fig{fig1}. The OSE substructure is found charge neutral,
293:  indicating there are no charging and polaron effects
294:  below the saturation voltage.
295: %
296:  The abrupt increase of current near the threshold
297:  $V_{\rm th}$ at 3.7 V should therefore be interpreted with
298:  an alternative mechanism.
299:   First of all, we shall show that the observed
300:  abrupt current change relates to an I-M transition.
301: %
302:  Presented in \Fig{fig2} are the changes of
303:  transmission coefficient $T(E)$ and lattice distortion $y_n$
304:  around the threshold voltage.
305:  Evidently, the Peierls phase of the OSE substructure
306:  is largely eliminated near $V_{\rm th}$.
307:  The key features in \Fig{fig2} include
308:  (1) the rapid increasing of transmission coefficient in
309:  the window between the indicated $\mu_{\rm L}$
310:  and $\mu_{\rm R}$ in \Fig{fig2}(a);
311:  (2) the distinct reducing of the lattice distortion
312:  or single-double bond alternation [\Fig{fig2}(b)];
313:  and also (3) the vanishing of the energy gap
314:  and the merging of valence and conductance bands,
315:  which will be explained later, cf.\ \Fig{fig3}(a).
316:  The above features all indicate an I-M transition
317:  around $V_{\rm th}$.
318: %
319:  When the bias sweeps down, the Peierls
320:  phase, as indicated by the lattice distortion
321:  in \Fig{fig2}(b) (see also \Fig{fig3} for energy gap),
322:  is reformed resulting in a metal-insulator (M-I)
323:  transition around 2.8 V.
324: %
325:  The delayed M-I transition, in comparing with
326:  the I-M transition voltage, is due to
327:  the lattice distortion reorganization,
328:  resulting in the hysteretic loop ($\sim 1$ V)
329:  in the $I$-$V$ curve in \Fig{fig1}.
330: 
331: %\input{figcap2}
332: 
333:   To further elucidate the bias-induced
334:  I-M/M-I transition mechanism, we also calculate the
335:  `reduced' energy levels and wave functions of the OSE substructure
336:  by diagonalizing its Hamiltonian. It consists of \Eq{SSH}, with
337:  the nonequilibrium lattice parameters being
338:  evaluated in the self-consistent
339:  manner as described earlier.
340:  The Hamiltonian contains also the
341:  electric potential variation in the OSE structure,
342:  which had also been determined self-consistently.
343:  The informations on this Hamiltonian are
344:  shown to be able to reflect the bias-induced intrinsic
345:  changes of the OSE in the device.
346: 
347:  The resulting energy levels from the
348:  aforementioned self-consistent Hamiltonian, as they vary
349:  with the up-sweeping and down-sweeping bias voltage,
350:  are reported in \Fig{fig3}, in which only
351:  six of them, HOMO-2, $\cdots$, LUMO+2, are examined, since they
352:  dominate most of the properties of the OSE during bias sweeping.
353:  The most striking feature of \Fig{fig3} is the bias-induced energy
354:  levels crossover. It occurs twice in the bias sweep-up
355:  [\Fig{fig3}(a)] but only once in the sweep-down direction
356:  [\Fig{fig3}(b)]. The first crossover in \Fig{fig3}(a) does not
357:  increase the current. Detailed analysis finds it due to the fact
358:  that the corresponding HOMO and LUMO states are localized near the
359:  metal-OSE interfaces by the surface vibration modes (not shown here
360:  due to space limitation). Note that the HOMO/LUMO here is a reduced state
361:  in the presence of electrodes and lattice.
362:  Unlike the first crossover, the second one, shown by the
363:  dual-crossing between the HOMO and LUMO+1
364:  and between the LUMO and HOMO-1 as indicated
365:  by the arrows in \Fig{fig3}(a) around $V_{\rm th}$,
366:  does increase the current, as it affects the system remarkably.
367:  It combines the
368:  valence and conductance bands of the OSE to form a single metallic
369:  band structure without energy gap;
370:  and meanwhile it also {\it completely
371:  delocalizes} the wave functions
372:  of those six levels over the whole OSE substructure
373:  The gap-free metallic band structure
374:  and the delocalized wave functions,
375:  together with the uniform valence bonds [cf.\ \Fig{fig2}(b)]
376:  around the second crossover,
377:  are highly advantageous to the transport of $\pi$-electrons.
378:  They lead to the I-M transition of the organic electronics
379:  with rapidly increasing current around the $V_{\rm th}$.
380: 
381:  In a similar but reversible manner,
382:  when the bias sweeps down the energy levels
383:  crossover at $V \sim 2.8$\,V
384:  re-opens the band gap [\Fig{fig3}(b)],
385:  re-localizes the electronic
386:  wave functions, and re-dimerizes the lattice displacements of the OSE.
387:  As results, the OSE substructure transforms
388:  from the metallic back to the insulating state (M-I transition)
389:  with the current sudden drop following
390:  a near-linear decrease (cf.\ \Fig{fig1}).
391: 
392:   Some experimental evidences of the bias-induced I-M transition
393:  have been reported in large size organic electronics.
394:  One is the bias-driven high-to-low resistive state
395:  transition in the
396:  alkali-7,7,8,8-tetracyanoquinodimethane (K-TCNQ) single crystals,
397:  a quasi-one dimensional organic charge-transfer complex
398:  (Mott-Peierls insulator) \cite{Kum991645}.
399:  The I-M transition occurs in this system
400:  at about several hundred of volt in low temperature, where
401:  a metallic path has been visualized with a microscope.
402: %
403:  The present study may also shed some light on understanding
404:   the experimental observations in Ref.\ \onlinecite{Mat0512450}.
405:  These include the conductance switching, hysteretic loop,
406:  and `dimerization' reduction in the TTF-based dimeric
407:   donor salt; cf.\ \Fig{fig1} here versus
408:   the Fig.~2 of Ref.\ \onlinecite{Mat0512450}.
409:   The detailed comparison between the
410:   theory and experiments will be published
411:  elsewhere.
412: 
413:  In summary, we have investigated the
414:  bias-induced I-M transition
415:  in organic electronics devices and
416:  explained it with the energy
417:  levels crossover which diminishes the Peierls phase and
418:  delocalizes the electron states
419:  at the threshold voltage.
420:  The bias-triggered intrinsic large on-off
421:  ratio in this work ($\gtrsim$ 20:1)
422:  is experimental controllable. Our theory study reproduces a
423:  range of experimental observations, such as
424:  the I-M phase transition \cite{Kum991645,Mat0512450},
425:  the hysteretic conductance switching \cite{Sim6777,Boz04607}.
426: 
427: 
428: Support from the National Natural Science Foundation of China
429: (Grants No.~10604037 and 10474056) and the Research Grants Council
430: of the Hong Kong Government (605105) is gratefully acknowledged.
431: 
432: \begin{thebibliography}{20}
433: 
434: \bibitem{Che991550}
435: J.~Chen, M.~A. Reed, A.~M. Rawlett, and J.~M. Tour,
436: \newblock Science {\bf 286}, 1550 (1999).
437: 
438: \bibitem{Don012303}
439: Z.~J. Donhauser, B.~A. Mantooth, K.~F. Kelly, L.~A. Bumm, J.~D.
440: Monnell, J.~J.
441:   Stapleton, D.~W.~P. Jr., A.~M. Rawlett, D.~L. Allara, J.~M. Tour, and P.~S.
442:   Weiss,
443: \newblock Science {\bf 292}, 2303 (2001).
444: 
445: \bibitem{Col001172}
446: C.~P. Collier, G.~Mattersteig, E.~W. Wong, Y.~Luo, K.~Beverly,
447: J.~Sampaio,
448:   F.~M. Raymo, J.~F. Stoddart, and J.~R. Heath,
449: \newblock Science {\bf 289}, 1172 (2000).
450: 
451: \bibitem{Kra022927}
452: I.~Kratochvilova, M.~Kocirik, A.~Zambova, J.~Mbindyo, T.~E.
453: Malloukc, and T.~S.
454:   Mayer,
455: \newblock J. Mater. Chem. {\bf 12}, 2927 (2002).
456: 
457: \bibitem{Raw03377}
458: A.~M. Rawlett, T.~J. Hopson, I.~Amlani, R.~Zhang, J.~Tresek, L.~A.
459: Nagahara,
460:   R.~K. Tsui, and H.~Goronkin,
461: \newblock Nanotechnology {\bf 14}, 377 (2003).
462: 
463: \bibitem{Ree013735}
464: M.~A. Reed, J.~Chen, A.~M. Rawlett, D.~W. Price, and J.~M. Tour,
465: \newblock Appl. Phys. Lett. {\bf 78}, 3735 (2001).
466: 
467: \bibitem{Ser03556}
468: R.~F. Service,
469: \newblock Science {\bf 302}, 556 (2003).
470: 
471: \bibitem{Gal05125}
472: M.~Galperin, M.~A. Ratner, and A.~Nitzan,
473: \newblock Nano Lett. {\bf 5}, 125 (2005).
474: 
475: \bibitem{Wei0682}
476: J.~H. Wei, S.~J. Xie, L.~M. Mei, J.~Berakdar, and Y.~J. Yan,
477: \newblock New J. Phys. {\bf 8}, 82 (2006).
478: 
479: \bibitem{Blu05167}
480: A.~S. Blum, J.~G. Kushmerick, D.~P. Long, C.~H. Patterson, J.~C.
481: Yang, J.~C.
482:   Henderson, Y.~X. Yao, J.~M. Tour, R.~Shashidhar, and B.~R. Ratna,
483: \newblock Nature Mater. {\bf 4}, 167 (2005).
484: 
485: \bibitem{Kum991645}
486: R.~Kumai, Y.~Okimoto, and Y.~Tokura,
487: \newblock Science {\bf 284}, 1645 (1999).
488: 
489: \bibitem{Mat0512450}
490: M.~M. Matsushita and T.~Sugawara,
491: \newblock J. Am. Chem. Soc. {\bf 127}, 12450 (2005).
492: 
493: \bibitem{Pei55}
494: R.~Peierls,
495: \newblock {\em Quantum Theory of Solids},
496: \newblock Oxford University Press, Oxford, 1955.
497: 
498: \bibitem{Su802099}
499: W.~P. Su, J.~R. Schrieffer, and A.~J. Heeger,
500: \newblock Phys. Rev. B {\bf 22}, 2099 (1980).
501: 
502: \bibitem{Bra02165401}
503: M.~Brandbyge, J.~L. Mozos, P.~Ordej\'on, J.~Taylor, and
504: K.~Stokbro,
505: \newblock Phys. Rev. B {\bf 65}, 165401 (2002).
506: 
507: \bibitem{Kel641515}
508: L.~V. Keldysh,
509: \newblock Zh.\ Eksp.\ Teor.\ Fiz. {\bf 47}, 1515 (1964),
510: \newblock [Sov.~Phys.~JETP {\bf 20}, 1018 (1965)].
511: 
512: \bibitem{Wei0508417}
513: J.~H. Wei, S.~J. Xie, L.~M. Mei, and Y.~Yan,
514: \newblock Organic Electronics, DOI: 10.1016/j.orgel.2007.03.002
515: 
516: 
517: \bibitem{He0663}
518: J.~L. He, B.~Chen, A.~K. Flatt, J.~J. Stephenson, C.~D. Doyle, and
519: J.~M. Tour,
520: \newblock Nature Mater. {\bf 5}, 63 (2006).
521: 
522: \bibitem{Sim6777}
523: J.~G. Simmonsa and R.~R. Verderbe,
524: \newblock Proc. Roy. Soc. A. {\bf 301}, 77 (1967).
525: 
526: \bibitem{Boz04607}
527: L.~D. Bozano, B.~W. Kean, V.~R. Deline, J.~R. Salem, and J.~C.
528: Scotta,
529: \newblock Appl. Phys. Lett. {\bf 84}, 607 (2004).
530: 
531: \end{thebibliography}
532: 
533: 
534: 
535: \clearpage
536: 
537: \begin{figure}
538:  \includegraphics[width=3.in]{fig1.eps}
539:  \caption{The hysteretic current as the function of sweeping bias
540:  of the model metal/OSE/metal electronics at $T=11$ K.
541:  The size of OSE is $N_o=40$ and the linear voltage
542:  sweep rate is at 0.1 V/sec.
543:  The insert is the threshold voltage $V_{\rm th}$
544:  as a function of the OSE chain
545:  length $N_o$.}
546:  \label{fig1}
547:  \end{figure}
548: 
549:  \begin{figure}
550:  \includegraphics[width=3. in]{fig2.eps}
551:  \caption{(color online) High (solid lines) and low (dash-dot lines) conductance
552:  states triggered  by the up-sweeping bias.
553:  (a) Transmission coefficient $T(E)$. The vertical
554:  dot lines indicate the window between
555:  $\mu_{\rm L}$ and $\mu_{\rm R}$ for the charge
556:  transmission; (b) Lattice distortion $y_n\equiv u_{n+1}-u_n$
557:  of the OSE substructure.}
558:  \label{fig2}
559:  \end{figure}
560: 
561: 
562:  \begin{figure}
563:  \includegraphics[width=3. in]{fig3.eps}
564:  \caption{(color online) The voltage-evolution of six energy levels
565:  of the OSE (HOMO-2, $\cdots$, LUMO+2) for up-sweeping (a) and
566:  down-sweeping bias (b). The arrows indicate the crossover of
567:  energy levels.}
568:  \label{fig3}
569:  \end{figure}
570: 
571: \end{document}
572: )
573: