1: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
2: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
3:
4:
5: % Some other (several out of many) possibilities
6: %\documentclass[preprint,aps]{revtex4}
7: %\documentclass[preprint,aps,draft]{revtex4}
8: %\documentclass[prb]{revtex4}% Physical Review B
9:
10: \usepackage{graphicx}% Include figure files
11: %\usepackage{dcolumn}% Align table columns on decimal point
12: %\usepackage{bm}% bold math
13:
14: %\nofiles
15:
16: \begin{document}
17:
18: \title{Cotunneling and one-dimensional localization in individual single-wall carbon nanotubes}
19:
20: \author{B. Gao,$^1$ D.C. Glattli,$^{1,2}$ B. Pla\c{c}ais,$^1$ A. Bachtold$^{1,3*}$}
21:
22: \address{
23: $^1$ Laboratoire Pierre Aigrain, Ecole Normale Sup\'{e}rieure, 24
24: rue Lhomond, 75231 Paris 05, France. $^2$ SPEC, CEA Saclay,
25: F-91191 Gif-sur-Yvette, France, $^3$ ICN and CNM-CSIC, Campus
26: UABarcelona, E-08193 Bellaterra, Spain.}
27:
28:
29:
30: \begin{abstract}
31: We report on the temperature dependence of the intrinsic
32: resistance of long individual disordered single-wall carbon
33: nanotubes. The resistance grows dramatically as the temperature is
34: reduced, and the functional form is consistent with an activated
35: behavior. These results are described by Coulomb blockade along a
36: series of quantum dots. We occasionally observe a kink in the
37: activated behavior that reflects the change of the activation
38: energy as the temperature range is changed. This is attributed to
39: charge hopping events between non-adjacent quantum dots, which is
40: possible through cotunneling processes.
41:
42:
43: \end{abstract}
44:
45: \vspace{.3cm} \pacs{73.63.Fg, 73.20.Fz, 73.23.Hk}
46:
47: \date{ \today}
48: \maketitle
49:
50: Single-wall carbon nanotubes (SWNT) are an excellent system to
51: study one-dimensional (1-d) transport. In particular, the effect
52: of disorder in 1-d is very pronounced; current lines have to
53: follow the wire and cannot go round impurity centers. As the
54: transmission of impurity centers becomes low enough, the 1-d wire
55: is divided in a series of quantum dots. The conduction is then
56: thermally activated $R(T) \approx \textmd{exp} (T^{-1})$
57: \cite{Ruzin,Staring,Chandrasekhar,Bezryadin}.
58:
59: Measurements on 2-d or 3-d arrays of quantum dots can show a
60: slower than thermally activated dependence of the conduction $R(T)
61: \approx \textmd{exp} (T^{-0.5})$ \cite{Beverly,Yu}. This has been
62: recently attributed to cotunneling processes, which allow charge
63: transfer between non-adjacent quantum dots \cite{Tran,Feigelman,
64: Beloborodov}. Indeed, cotunneling transport in a series of quantum
65: dots is analogous to variable-range hopping (VRH)
66: \cite{Shklovskii}. Charges try to find hopping events with the
67: lowest activation energy and the shortest hopping distance. The
68: slower than thermally activated dependence of the conduction is
69: then a result of successive thermally activated curves with the
70: activation energy that decreases as the temperature is reduced.
71: However, such a succession of activated curves remain to be
72: observed.
73:
74: Localization experiments have been carried on nanotube films or
75: individual SWNTs contacted to microfabricated electrodes, but
76: tube-tube junctions and tube-electrode interfaces make the
77: analysis difficult \cite{Bezryadin,Shea,Jang,Fuhrer,Vavro}. In our
78: experiments, the intrinsic resistance of disordered SWNTs is
79: measured in a four-point configuration \cite{Gao2}. The intrinsic
80: resistance is found to be thermally activated. As the gate voltage
81: ($V_g$) is swept, we observe Coulomb blockade oscillations that
82: can be rather regular in some cases. These measurements are
83: consistent with a series of quantum dots that are typically
84: $\gtrsim 10$~nm long. Importantly, we also observe kinks in the
85: activated behavior of $R(T)$ that suggest the change of the
86: activation energy as the temperature range is varied. These kinks
87: are attributed to cotunneling processes.
88:
89: \begin{figure}
90: \includegraphics{fig1}
91: \caption{
92: Four-point resistance. (a) Device schematic. (b-h) $R_{4pt}$ as a
93: function of temperature. When the value of $V_g$ is not indicated,
94: $V_g=0$. Device 3b is the same device as Device 3, but it has been
95: measured one month before. Microscopic changes might have been
96: occurred in between. }
97: \end{figure}
98:
99: The fabrication of SWNT devices for four-point measurements has
100: been described in Ref. \cite{Gao2}. Briefly, $\sim1$ nm diameter
101: SWNTs grown by laser-ablation \cite{Thess} are selected with an
102: atomic force microscopy (AFM). Noninvasive voltage electrodes are
103: defined by positioning two MWNTs above the SWNT using AFM
104: manipulation. Cr/Au electrodes are patterned for electric
105: connection using electron-beam lithography (Fig. 1(a)).
106: Characteristics of the devices are summarized in Table 1.
107:
108: \begin{table}
109: \includegraphics{Tab}
110: \caption{
111: Device characteristics. $L$ is the length between the MWNTs.
112: $L_{dot}$ is calculated from $E_0$ extracted at $V_g=0$. SC =
113: semiconducting tube with the threshold voltage at $\sim 40$ V.
114: }
115: \end{table}
116:
117: The four-point resistance $R_{4pt}$ of some SWNTs is particularly
118: large $> 100$ k$\Omega$ at 300~K. The nature of the scattering
119: centers responsible for this resistance is at present not
120: understood. Figure 1(b) shows the temperature dependence of
121: $R_{4pt}$ of one of those SWNTs (device 1). The curve is quite
122: flat at high $T$, while the resistance increases a lot below
123: 100~K. The high-temperature resistance allows the estimation of
124: the elastic mean-free path $L_e$. Using $R_{4pt}=h/4e^2\cdot
125: L/L_e$ with $L$ the length between the MWNTs \cite{Gao2}, we get
126: $L_e=18$~nm.
127:
128: For comparison, we also show a device that is significantly less
129: resistive at 300~K, $R_{4pt}=$12~k$\Omega$. The $R_{4pt}(T)$
130: variation is much less pronounced. This is consistent with
131: previously reported works on two-point, low ohmic SWNT devices,
132: where the $R_{2pt}(T)$ dependence is weak
133: \cite{Nygard,Liang,Kong}. For two-point devices with a large
134: resistance, the resistance has been reported to strongly grow as
135: $T$ goes to zero, which is usually associated to the change of the
136: contact resistance \cite{Bockrath}. In our case, the four-point
137: technique allows to separate the intrinsic and contact
138: resistances.
139:
140: Figure 1(c) shows that the above measurement in the highly
141: diffusive tube is consistent with an activated behavior of the
142: resistance
143:
144: %====================================================================
145: \begin{equation}
146: R_{4pt}=R_0\exp\frac{E_0}{kT}
147: \end{equation}
148: %====================================================================
149: with $E_0$ the activation energy. This dependency is observed in
150: other devices (see Fig. 1(d-f)). Similar $R(T)$ behaviors have
151: been reported for disordered wires microfabricated in
152: semiconductors \cite{Khavin,Kastner}.
153:
154: Figures 1(f-h) show $R(T)$ measurements on other tubes. Some of
155: them deviate from the standard activated behavior. However, these
156: measurements can be described by successive exponential functions
157: with different activation energies, giving rise to kinks.
158: Interestingly, Fig. 1(f) shows that those two exponential
159: functions can merge in a single one on varying the gate voltage,
160: which is applied on the back side of the Si wafer. Overall, these
161: measurements suggest that the activation energy depends on the
162: temperature range and the gate voltage.
163:
164: Further insight into transport properties is obtained by studying
165: the high-voltage regime. Figure 2 shows that the differential
166: $R_{4pt}$ is lowered as $V_{4pt}$ increases, and that the
167: dependence can be fitted with
168:
169: %====================================================================
170: \begin{equation}
171: R_{4pt}=R_0\exp\frac{E_0-\alpha eV_{4pt}/2}{kT}
172: \end{equation}
173: %====================================================================
174:
175: This suggests that an increase in the voltage reduces the
176: activation energy. An important point is that the slope deduced
177: from Fig. 2 gives $\alpha$ in Eq. (2) below unity. This means that
178: more than one energy barrier has to be overcome along the tube. A
179: rough estimate of the number of barriers $N$ can be made by taking
180: $N=1/\alpha$, which assumes identical barriers \cite{Khavin}. In
181: this way we obtain $N \lesssim 20$.
182:
183:
184:
185: \begin{figure}
186: \includegraphics{fig2}
187: \caption{
188: Four-point differential resistance as a function of $V_{4pt}$ for
189: Device 1 at 35, 45, 55, and 60~K.}
190: \end{figure}
191:
192:
193:
194: Fig. 3(a) shows the effect of the gate voltage, which controls the
195: position of the Fermi level in the tube. Large fluctuations of
196: $R_{4pt}(V_g)$ develop at low $T$ that look random
197: \cite{Kastner,Fowler}. At first sight, this may question the
198: activation behavior of $R_{4pt}(T)$ and the kinks discussed above.
199: However, Fig. 3(b) shows that a curve similar to Fig. 1(f)
200: ($V_g=0$) is found by $V_g$ averaging $R_{4pt}(T)$. Moreover,
201: similar dependencies are observed, albeit with different
202: activation energies, for the minima and maxima of $R_{4pt}(V_g)$
203: as a function of $T$. This illustrates the robustness of the
204: activation behavior and the kink for a nanotube with a given Fermi
205: level.
206:
207:
208:
209:
210: \begin{figure}
211: \includegraphics{fig3}
212: \caption{
213: Four-point resistance as a function of the gate voltage. (a)
214: $R_{4pt}(V_{g})$ for Device 3b at 18, 25, 31, 39, 49, 60, 74, 96,
215: 125, 158, and 191~K. (b) $R_{4pt}(T)$ averaged over $V_g$ between
216: -0.3 and 0.3~V and taken at different conductance maxima
217: (m1@0.09V, m2@0.17V, m3@-0.25V, m4@-0.05V). (c) Two-point
218: differential conductance as a function of $V_{g}$ and $V_{2pt}$ at
219: 20~K. The same measurement with $G_{4pt}$ is very noisy.
220: $R_{4pt}(V_{g})$ and $R_{2pt}(V_{g})$ in the linear regime show
221: the same features at 20~K. Figures 1(f), 3(a) and 3(c) are taken
222: in three cooling runs. (d,e) $G_{4pt}(V_{g})$ for device 4. }
223: \end{figure}
224:
225: While these fluctuations look random, oscillations can be found
226: that are quite regular within restricted $V_g$ ranges
227: \cite{Field}. Fig. 3(c) shows 10 successive oscillations. Note
228: that series of regular oscillations can be found at other $V_g$
229: ranges, and the period is then identical. Interestingly, Fig.
230: 3(d,e) show that the period can change as the temperature is
231: modified. New oscillations can appear at lower $T$ that have a
232: shorter period.
233:
234: We now discuss possible origins for the activated behavior of the
235: resistance. One possible mechanism is the Schottky barrier at the
236: interface between a metal electrode and a semiconducting nanotube
237: \cite{Martel}. However, we also observe the activated behavior in
238: metal tubes, which have no Schottky barriers. Moreover, the
239: four-point technique is aimed to avoid contributions from the
240: contacts \cite{Gao2}. Another mechanism is thus needed to account
241: for the results.
242:
243: The fluctuations of $R_{4pt}(V_g)$ and the $R_{4pt}(T)$ dependence
244: may be attributed to universal conductance fluctuations and weak
245: localization. However, the variations of $R_{4pt}$ are much larger
246: than $h/e^2$, so that the results cannot simply originate from
247: interference corrections.
248:
249: Strong localization (SL) is expected for highly diffusive systems
250: \cite{Shklovskii}. This theory has been used to explain
251: exponential length dependencies of the resistance measured in
252: nanotubes \cite{Pablo,Cumings,Gomez}. SL occurs when the
253: phase-coherence length $L_{\phi}$ becomes longer than the
254: localization length $L_{loc}$. This is equivalent to when the
255: width of the coherent states, $\hbar v_F/L_{\phi}$, becomes
256: smaller than the energy separation between the localized states.
257: The localized states are usually regarded as randomly distributed
258: in space and energy (see Fig. 4(a)) \cite{Imry,Fogler}. Irregular
259: oscillations of $R_{4pt}(V_g)$ are expected, which is in
260: opposition to our results.
261:
262:
263: \begin{figure}
264: \includegraphics{fig4}
265: \caption{
266: Schematics of localized states along the nanotube. (a) States are
267: randomly distributed. (b) Strong barriers that define quantum
268: dots. (c) Proposed process to account for the kinks in Fig. 1
269: (f,g). Arrows represent hopping paths. }
270: \end{figure}
271:
272:
273: We now look at an alternative distribution of localized states as
274: schematized in Fig. 4(b). The tube is here divided in segments
275: separated by highly resistive scattering centers. The segment
276: lengths and therefore the energy separations can be different. At
277: high enough temperatures, levels are thermally smeared out except
278: for the shortest segment that has the largest level separations.
279: Oscillations of $R_{4pt}(V_g)$ are then regular, and the period is
280: large. At lower temperature, shorter periods arise from longer
281: segments, which agrees with experiments.
282:
283: So far, the Coulomb interaction between electrons has not been
284: taken into account. However, the charging energy $E_c$ of a single
285: nanotube quantum dot is known to be larger than the level spacing
286: $\Delta E$ due to the geometrical confinement of the electron
287: wave. $\Delta E \approx 0.5 $ meV$\cdot \mu$m and the charging
288: energy for a tube dot connected to two tube leads is roughly $E_c
289: \approx 1.4 $ meV$\cdot \mu$m \cite{Bozovic}. This suggests that
290: the separation in energy between the localized states in Fig. 4(b)
291: is given by the charging energy.
292:
293: Localization related to Coulomb blockade through multiple quantum
294: dots \cite{Ruzin,Staring,Chandrasekhar,Bezryadin} bears a lot in
295: common with the standard hopping model of the strong localization
296: theory \cite{Imry,Kastner,Fowler,Fogler}. Series of aperiodic
297: conductance oscillations are expected. Contrary to the SL regime,
298: however, quasi-periodic oscillations are also occasionally
299: predicted, in agreement with experiments. In addition the
300: resistance is expected to be thermally activated, which again
301: agrees with experiments. The activation energy is given by the dot
302: with the level that lies the furthest away from the Fermi level.
303: It may also be the largest separation of energy levels located in
304: neighboring dots. Thus, $E_0$ is expected to be gate voltage
305: dependent, consistent with our experimental findings.
306:
307: We here estimate the size of the dots. The activation energy $E_0$
308: is roughly $0.5E_c$ of the shortest dot. $E_0$ is 11.5~meV for
309: device 4 at high $T$. Using $E_c \approx 1.4 $ meV$\cdot \mu$m, we
310: get a dot length of $\sim 60$~nm. Another possibility for this
311: estimation is to use the 625~meV period of the $R_{4pt}(V_g)$
312: oscillations at high $T$ (Fig. 3(e)). Indeed, $\Delta V_g\approx
313: 12.5$~meV$\cdot \mu$m when looking Ref.\cite{Postma,Bozovic} for a
314: tube dot connected to two tube leads. This gives $\sim 20$~nm.
315: Note that $E_c$ cannot be estimated from the diamond height in
316: Fig. 3(c) since several dots lie in series. Finally, we obtain
317: $\sim 70$~nm by dividing the tube length by the dot number
318: obtained in Fig. 2. Those 3 estimations point all to quantum dot
319: lengths of a few 10~nm.
320:
321: Table 1 gives the dot length of the other samples, estimated from
322: $E_0$. Dot lengths are slightly longer than the elastic length
323: $L_e$ determined at 300~K. $L_e$ corresponds to the separation
324: between scatterers when transmissions are 0.5. The barriers that
325: define the quantum dots thus have a transmission $\lesssim 0.5$,
326: which corresponds to a resistance $\gtrsim 6.5$~k$\Omega$. This is
327: consistent with the occurrence of Coulomb blockade since the
328: barrier resistance has to be larger than a few k$\Omega$s.
329:
330:
331: Having shown that the activated behavior of $R_{4pt}(T)$
332: originates from a series of quantum dots, we now turn our
333: attention to the kinks (Fig. 1(f-h)). This may simply come from
334: two thermally activated resistances that lie in series. However,
335: the activation energy would be higher at lower $T$, in opposition
336: to the measurements. Another mechanism is needed to describe the
337: kinks.
338:
339: We propose that the kink is related to a mechanism that is
340: borrowed from the theory of variable range hopping
341: \cite{Shklovskii}, see Fig. 4(c). Electrons hop to the neighboring
342: quantum dot as indicated by the arrow 1. At lower $T$ it pays to
343: make the hop 2 to the second nearest quantum dot. The activation
344: energy is given by the level separation, which is thus reduced.
345: This is in agreement with the experiments.
346:
347: In the VHR theory such hops are possible thanks to the tunneling
348: process. However, the tunnel probability is here dramatically low
349: since the second nearest dot is a few tens of nanometers far.
350: Another mechanism for the charge transfer between nonadjacent
351: quantum dots is needed to account for the results.
352:
353: A possible mechanism is that the charge motion between two
354: nonadjacent dots occurs through cotunneling events
355: \cite{Tran,Feigelman, Beloborodov}. Cotunneling, which involves
356: the simultaneous tunneling of two or more electrons, transfers the
357: charge via a virtual state. This gives rise to current even when
358: the electron transport is Coulomb blockaded \cite{Pasquier}. A
359: cotunneling event is called inelastic when the quantum dot is left
360: in an excited state, and the event is otherwise called elastic.
361: For an individual quantum dot contacted to two leads, the
362: conductance contribution of elastic cotunneling is temperature
363: independent, while the contribution of inelastic cotunneling
364: scales as $T^2$.
365:
366: Cotunneling in a series of quantum dots has been recently
367: calculated \cite{Feigelman, Beloborodov}. An energy reservoir
368: supplied by for e.g. phonons is required since $\epsilon _i$ the
369: energy of the initial state is most often different than $\epsilon
370: _f$ the energy of the final state (see hop 2 in Fig. 4(c)). The
371: resistance contribution between those two states is
372: \cite{Feigelman}
373:
374: %====================================================================
375: \begin{equation}
376: R \propto R_0^N\exp\frac{\textmd{max}(|\epsilon _i-\epsilon
377: _f|,|\epsilon _i-\mu|,|\epsilon _f-\mu|)}{kT}
378: \end{equation}
379: %====================================================================
380: with $\mu$ the Fermi level and $N$ the number of dots between the
381: initial and the final states. $R_0=A_1E_c/(g\Delta E)$ for elastic
382: cotunneling and $R_0=A_2N^2E_c^2/(g(\epsilon _i-\epsilon _f)^2)$
383: for inelastic cotunneling with $g=Gh/e^2$ the average
384: dimensionless conductance of a barrier between two dots and $A_1$
385: and $A_2$ numerical constants of the order of unity. The coulomb
386: repulsion term between the dots $i$ and $f$ is here neglected for
387: simplicity. The prefactor $R_0^N$ grows as $N$ the number of
388: involved barriers gets larger. At high temperature, the hopping
389: process between two adjacent dots dominates transport and the
390: prefactor is low (hop 1 in Fig. 4(c)). As the temperature is
391: reduced, the exponential term grows a lot. It then pays to make
392: the hop between non-adjacent dots when the activation energy is
393: lower (hop 2 in Fig. 4(c)). This is consistent with the kinks
394: observed in Fig. 1.
395:
396: The temperature $T^*$ of the first kink is expected to be around
397: $kT^* \simeq E_0^{above}-E_0^{below}$ with $E_0^{below}$ and
398: $E_0^{above}$ the activation energies below and above $T^*$. This
399: can be obtained from Eq. 3 taking into account that
400: $N^{below}-N^{above}=1$ and that $\ln R_0$ is of the order of
401: unity. This relation is consistent with the experiments. For
402: example, $E_0^{above}-E_0^{below}=14$ meV in Fig. 1(f) for $V_g=0$
403: while $kT^*=6$ meV.
404:
405: We have seen that cotunneling processes allow a slower than
406: thermally activated dependence of the conduction. The main
407: contribution of the conduction comes from one (or a few) quantum
408: dot. The energy levels are randomly positioned in energy, so that
409: we cannot expect a specific functional form for the slower than
410: activated dependence measured here.
411:
412: In conclusion, we have shown that the intrinsic resistance of
413: strongly disordered SWNTs is thermally activated. This is due to
414: Coulomb blockade in a series of $\gtrsim 10$~nm long quantum dots
415: lying along the tube. The activation energy is found to change as
416: the temperature range is changed. We attribute this result to
417: cotunneling processes. Disordered SWNTs form an interesting system
418: for future studies on one-dimensional localization. For example,
419: studies on longer tubes will be investigated to reach the 1-d
420: variable range hopping regime \cite{Fogler2}.
421:
422:
423: We thank C. Delalande for support, L. Forro for MWNTs, R. Smalley
424: for laser-ablation SWNTs, and P. McEuen, J.L. Pichard, M. Fogler
425: and M. Feigelman for discussions. LPA is CNRS-UMR8551 associated
426: to Paris 6 and 7. The research has been supported by ACN, Sesame.
427:
428:
429: $^{*}$ corresponding author: adrian.bachtold@cnm.es
430:
431:
432: \begin{references}
433:
434: \bibitem{Ruzin}
435: I.M. Ruzin, V. Chandrasekhar, E.I. Levin, and L.I. Glazman, Phys.
436: Rev. B \textbf{45}, 13469 (1992).
437:
438: \bibitem{Staring}
439: A.A.M. Staring, H. van Houten, and C.W.J. Beenakker, Phys. Rev. B
440: \textbf{45}, 9222 (1992).
441:
442: \bibitem{Chandrasekhar}
443: V. Chandrasekhar, Z. Ovadyahu, and R. A. Webb, Phys. Rev. Lett.
444: \textbf{67}, 2862 (1991).
445:
446: \bibitem{Bezryadin}
447: A. Bezryadin, A.R.M. Verschueren,S.J. Tans, and C. Dekker, Phys.
448: Rev. Lett. \textbf{80}, 4036 (1998).
449:
450: \bibitem{Beverly}
451: K.C. Beverly, J.F. Sampaio, J.R. Heath, J. Phys. Chem. B
452: \textbf{106}, 2131 (2002).
453:
454: \bibitem{Yu}
455: D. Yu, C. Wang, B. L. Wehrenberg, and P. Guyot-Sionnest, Phys.
456: Rev. Lett. \textbf{92}, 216802 (2004).
457:
458: \bibitem{Tran}
459: T.B. Tran, I.S. Beloborodov, X.M. Lin, T.P. Bigioni, V.M. Vinokur,
460: and H.M. Jaeger, Phys. Rev. Lett. \textbf{95}, 076806 (2005).
461:
462: \bibitem{Feigelman}
463: M.V. Feigel'man, A.S. Ioselevich, JETP Lett. \textbf{81}, 277
464: (2005).
465:
466: \bibitem{Beloborodov}
467: I.S. Beloborodov, A.V. Lopatin, V.M. Vinokur, Phys. Rev. B
468: \textbf{72}, 125121 (2005).
469:
470: \bibitem{Shklovskii}
471: B.I. Shklovskii and A.L. Efros, \textit{Electronic Properties of
472: Doped Semiconductors} (Springer, New York, 1984).
473:
474: \bibitem{Shea}
475: H.R. Shea, R. Martel, and P. Avouris, Phys. Rev. Lett.
476: \textbf{84}, 4441 (2000).
477:
478: \bibitem{Jang}
479: W. Y. Jang, N.N. Kulkarni, C.K. Shih, and Z. Yao , Appl. Phys.
480: Lett. \textbf{84}, 1177 (2004).
481:
482: \bibitem{Fuhrer}
483: M.S. Fuhrer, M.L. Cohen, A. Zettl and V. Crespi, Solid Stat
484: Commun. \textbf{109}, 105 (1998).
485:
486: \bibitem{Vavro}
487: J. Vavro, J.M. Kikkawa, and J.E. Fischer, Phys. Rev. B
488: \textbf{71}, 155410 (2005).
489:
490: \bibitem{Gao2}
491: B. Gao, Y.F. Chen, M.S. Fuhrer, D.C. Glattli, and A. Bachtold,
492: Phys. Rev. Lett. \textbf{95}, 196802 (2005).
493:
494: \bibitem{Thess}
495: A. Thess, R. Lee, P. Nikolaev, H. Dai, P. Petit, J. Robert, C. Xu,
496: Y.H. Lee, S.G. Kim, A.G. Rinzler, D.T. Colbert, G.E. Scuseria, D.
497: Tom\a'anek, J.E. Fischer, and R.E. Smalley , Science \textbf{273},
498: 483 (1996).
499:
500: \bibitem{Nygard}
501: J. Nygard, D.H. Cobden, P.E. Lindelof, Nature \textbf{408}, 6810
502: (2000).
503:
504: \bibitem{Liang}
505: W. Liang, M. Bockrath, D. Bozovic, J.H. Hafner, M. Tinkham and H.
506: Park, Nature \textbf{411}, 665 (2001).
507:
508: \bibitem{Kong}
509: J. Kong, E. Yenilmez, T.W. Tombler, W. Kim, and H. Dai, R.B.
510: Laughlin, L. Liu, C.S. Jayanthi, and S.Y. Wu, Phys. Rev. Lett.
511: \textbf{87}, 106801 (2001).
512:
513: \bibitem{Bockrath}
514: M. Bockrath, D.H. Cobden, J. Lu, A.G. Rinzler, R.E. Smalley, L.
515: Balents, and P.L. McEuen, Nature {\bf 397}, 598 (1999).
516:
517: \bibitem{Khavin}
518: Y.B. Khavin, M.E. Gershenson, and A.L. Bogdanov, Phys. Rev. B
519: \textbf{58}, 8009 (1998).
520:
521: \bibitem{Kastner}
522: M.A. Kastner, R.F. Kwasnick, and J.C. Licini, Phys. Rev. B
523: \textbf{36}, 8015 (1987).
524:
525: \bibitem{Fowler}
526: A.B. Fowler, A. Hartstein, and R.A. Webb, Phys. Rev. Lett.
527: \textbf{48}, 196 (1982).
528:
529: \bibitem{Field}
530: S.B. Field, M.A. Kastner, U. Meirav, and J.H.F. Scott-Thomas, D.A.
531: Antoniadis and H.I. Smith, S.J. Wind, Phys. Rev. B \textbf{42},
532: 3523 (1990).
533:
534: \bibitem{Martel}
535: R. Martel, V. Derycke, C. Lavoie, J. Appenzeller, K.K. Chan, J.
536: Tersoff, and Ph. Avouris, Phys. Rev. Lett. \textbf{87}, 256805
537: (2001).
538:
539: \bibitem{Pablo}
540: P.J. de Pablo, C. G\a'omez-Navarro, J. Colchero, P.A. Serena, J.
541: G\a'omez-Herrero, and A.M. Bar\a'o, Phys. Rev. Lett. \textbf{88},
542: 036804 (2002).
543:
544: \bibitem{Cumings}
545: J. Cumings, and A. Zettl, Phys. Rev. Lett. \textbf{93}, 086801
546: (2004).
547:
548: \bibitem{Gomez}
549: C. G\a'omez-Navarro, P.J. de Pablo, J.G\a'omez-Herrero, B.Biel,
550: F.J.Garcia-Vidal, A.Rubio, and F. Flores, Nature Mat., \textbf{4},
551: 534 (2005).
552:
553:
554: \bibitem{Imry}
555: Imry, \textit{Electronic Transport in Mesoscopic Systems}
556: (Cambridge University Press, Cambridge, 1997).
557:
558: \bibitem{Fogler}
559: M.M. Fogler and R.S. Kelley, Phys. Rev. Lett. \textbf{95}, 166604
560: (2005).
561:
562: \bibitem{Bozovic}
563: D. Bozovic, M. Bockrath, J.H. Hafner, C.M. Lieber, H. Park, and M.
564: Tinkham, Appl. Phys. Lett. \textbf{78}, 3693 (2001).
565:
566:
567: \bibitem{Postma}
568: H.W.C. Postma, T. Teepen, Z. Yao, M. Grifoni, C. Dekker, Science
569: \textbf{293}, 76 (2001).
570:
571:
572: \bibitem{Pasquier}
573: C. Pasquier, U. Meirav, F.I.B. Williams, and D.C. Glattli, Y. Jin
574: and B. Etienne, Phys. Rev. Lett. \textbf{70}, 69 (1993); D.V.
575: Averin and Y.V. Nazarov, \textit{ibid} \textbf{65}, 2446 (1990).
576:
577: \bibitem{Fogler2}
578: M.M. Fogler, S.V. Malinin, T. Nattermann, cond-mat/0602008.
579:
580: \end{references}
581:
582:
583:
584:
585:
586: \end{document}
587: