1: \documentstyle[prl,aps,twocolumn]{revtex}
2: %TCIDATA{OutputFilter=LATEX.DLL}
3: %TCIDATA{LastRevised=Wed Jan 17 09:39:46 2001}
4: %TCIDATA{<META NAME="GraphicsSave" CONTENT="32">}
5: %TCIDATA{Language=American English}
6: %TCIDATA{CSTFile=revtex.cst}
7:
8: \tolerance = 10000
9:
10: \begin{document}
11: \input epsf
12: \twocolumn[\hsize\textwidth\columnwidth\hsize\csname
13: @twocolumnfalse\endcsname
14:
15: \title{Direct link between Coulomb blockade and shot noise in a quantum
16: coherent structure}
17: \author{A. Levy Yeyati$^{1}$, A. Martin-Rodero$^{1}$, D. Esteve$^{2}$ and C.
18: Urbina$^{2}$}
19: \address{$^{1}$Departamento de F\'\i sica Te\'orica de la Materia
20: Condensada C-V. \\
21: Facultad de Ciencias, \\
22: Universidad Aut\'onoma de Madrid, E-28049 Madrid, Spain \\
23: $^{2}$Service de Physique de l'Etat Condens\'{e}, Commissariat \`{a} \\
24: l'Energie Atomique, Saclay, F-91191 Gif-sur-Yvette Cedex, France}
25: \maketitle
26:
27: \begin{abstract}
28: We analyze the current-voltage characteristic of a quantum conduction
29: channel coupled to an electromagnetic environment of arbitrary
30: frequency-dependent impedance. In the weak blockade regime
31: the correction to the ohmic behavior is directly related to the channel
32: current fluctuations
33: vanishing at perfect transmission in the same way as shot noise.
34: This relation can be generalized to describe the environmental Coulomb
35: blockade in a generic mesoscopic conductor coupled to an
36: external impedance, as the response of the latter to the current
37: fluctuations in the former.
38: \end{abstract}
39:
40: \pacs{PACS numbers: 73.63.-b, 73.23.Hk}
41: \vskip2pc]
42:
43: \narrowtext
44:
45: The way in which quantum mechanics affect the laws ruling electrical
46: circuits is presently understood only for some elementary situations.
47: For instance, in the case of a quantum coherent nanostructure connecting two
48: independent electron reservoirs with a voltage difference $V$, quantum
49: mechanics results in current fluctuations, the so-called shot noise,
50: which at low
51: frequency and zero temperature have a spectrum of the form
52: $S=2eVG_{0}\sum \tau _{i}(1-\tau _{i})$, where the $\{\tau _{i}\}$ are
53: the transmissions of the conduction channels and $G_{0}=2e^{2}/h$ is the
54: conductance quantum \cite{noise1}. The reduction of noise with increasing
55: transmission as $(1-\tau _{i})$ is a consequence of the Fermi statistics in
56: the reservoirs, a prediction which has been tested quantitatively in
57: different types of nanostructures \cite{exp-noise}. Another important
58: consequence of quantum mechanics is that Ohm's law does not apply when two
59: elements are connected in series, because each element is not simply voltage
60: biased. The phase across each element $\phi =\frac{e}{\hbar} \int v(t)dt$,
61: where $v(t)$ is the voltage drop,
62: develops quantum fluctuations, and the electrical properties of the
63: series-connection cannot be inferred from the conductance of the separate
64: elements. More generally, the phase differences add in series, and parallel
65: connected branches share the same phase, but the general rules to predict the
66: properties of the whole circuit from those of the constitutive
67: elements are not known, except for macroscopic electromagnetic impedances.
68: When a low transmissive nanostructure, i.e. one with negligible noise
69: reduction, is
70: connected in series with a macroscopic impedance $Z(\omega )$, the
71: conductance of the series circuit is suppressed at sufficiently low voltage
72: and temperature \cite{devoret}. This phenomenon, called environmental Coulomb
73: blockade, has been thoroughly investigated in small metallic tunnel
74: junctions \cite{junctions}. How is this phenomenon modified in the case of a
75: coherent structure whose transmissions $\tau _{i}$ approach unity? One might
76: speculate that a noiseless structure cannot be ``felt'' by the series
77: impedance $Z(\omega)$, and, reciprocally, should not be affected by its
78: presence. We show here that this naive reasoning which predicts the
79: restoration of Ohm's law at large transmission is correct, and more
80: precisely that Coulomb blockade is modified in the same way as shot noise.
81:
82: In this Letter we address the simple case of a single channel
83: quantum point contact with transmission $\tau $ connected in series
84: with a macroscopic impedance $Z(\omega )$.
85: We develop a theory of the environmental Coulomb blockade which
86: permits to analyze the current-voltage characteristic at arbitrary
87: transmission for a generic frequency-dependent impedance. We show that
88: in the limit of low impedance $Z\ll 1/G_{0}$ the blockade is intimately
89: connected with the fluctuations in the current through the channel.
90:
91: A single channel contact of arbitrary transmission between two
92: electrodes can be modeled with a simple Hamiltonian
93: resembling the usual tunneling Hamiltonian for a tunnel junction
94: \cite{cuevas}. The coupling to the environment can be then introduced in the
95: usual way \cite{devoret}, which leads to a model Hamiltonian
96: $\hat{H}=\hat{H}_{L}+\hat{H}_{R}+\hat{H}_{T}$, where $\hat{H}_{L,R}$
97: describe the uncoupled
98: left and right leads, characterized by flat densities of states $\rho
99: _{L,R}\simeq 1/\pi W$ and
100:
101: \begin{equation}
102: \hat{H}_{T}=\sum_{\sigma }T_{0}\hat{c}_{L\sigma }^{\dagger }\hat{c}_{R\sigma
103: }\hat{\Lambda}_{e}+\,\,h.c., \label{hamiltonian}
104: \end{equation}
105: describes the transfer of an electron between the leads in terms of a
106: hopping element $T_0$. The translation operator
107: $\hat{\Lambda}_{e}=e^{i\hat{\phi}}$ , where
108: $\hat{\phi}$ is the phase operator satisfying the commutation relation
109: $[\hat{Q},\hat{\phi}]=ie$, takes
110: into account the change in the charge $\hat{Q}$ of the
111: environment associated with the transfer process.
112: If the coupling to the environment is neglected the normal transmission of
113: this model is given by $\tau =4\beta /(1+\beta )^{2}$,
114: where $\beta =(T_{0}/W)^{2}$ \cite{cuevas}.
115:
116: We are interested in calculating the current through the channel under a
117: constant bias voltage $V$ in the presence of the environment. The current
118: operator within this model is given by
119:
120: \begin{equation}
121: \hat{I} = \frac{ie}{\hbar} \sum_{\sigma} T_0 \hat{c}^{\dagger}_{L\sigma}
122: \hat{c}_{R\sigma} \hat{\Lambda}_e - \,\, h.c. , \label{current}
123: \end{equation}
124:
125: In order to evaluate the mean current we use the Keldysh formalism \cite
126: {keldysh} which is suitable for calculating averages in a non-equilibrium
127: state. The mean current is formally given by
128:
129: \begin{equation}
130: <\hat{I}(t)> = <\hat{T}_c \left[ \hat{I}_I(t) \hat{S}_c(\infty,-\infty) %
131: \right]> ,
132: \end{equation}
133: where $\hat{T}_c$ is the chronological ordering operator along the Keldysh
134: contour, $\hat{I}_I$ is the current operator in the interaction
135: representation and $\hat{S}_c(\infty,-\infty)$ is the corresponding
136: evolution operator along the closed time contour. Introducing the series
137: expansion of $\hat{S}_c$ in terms of $\hat{H}_T$ and applying Wick theorem
138: we obtain a perturbative expansion for the evaluation of the current which
139: can be expressed in terms of Keldysh Green functions. The lowest order
140: diagrams within this theory are depicted in Fig. 1a. In these diagrams we
141: associate a full line with an arrow to the electron propagators for the
142: uncoupled leads (denoted below by $g^{\alpha,\beta}_{L,R}$), crosses
143: indicate hopping events, and wavy lines correspond to the environment
144: correlators given by \cite{comment}
145:
146: \begin{equation}
147: P^{\alpha,\beta}(t,t^{\prime}) = e^{J^{\alpha,\beta}(t,t^{\prime})} ,
148: \end{equation}
149: where $\alpha,\beta \equiv +,-$ indicate the branch on the Keldysh contour
150: for the two time arguments, and $J^{\alpha,\beta}$ are the phase correlation
151: functions
152: \[
153: J^{\alpha,\beta}(t,t^{\prime}) = <\hat{T}_c \left[\hat{\phi}(t_{\alpha})
154: \hat{\phi}(t^{\prime}_{\beta}) \right]> - <\hat{\phi}^2> .
155: \]
156:
157: As usual, we assume that the modes in the environment (or photon states) are
158: populated according to the equilibrium distribution at a given temperature.
159: In this case $J^{\alpha,\beta}(t,t^{\prime}) =
160: J^{\alpha,\beta}(t-t^{\prime}) $.
161:
162: The evaluation of the complete perturbative series is a formidable task. One
163: can, however, obtain useful results in the limit of weak impedance $Z \ll
164: 1/G_0$. In this limit $P^{\alpha,\beta}$ can be approximated by $1 +
165: J^{\alpha,\beta}$. To the first order in $J^{\alpha,\beta}$ one obtains the
166: family of diagrams depicted in Fig 1b, where the phase correlation functions
167: are represented by a dashed line. These diagrams correspond to single
168: ``photon'' processes and can be associated into four groups depending on the
169: types of hopping event (left to right or right to left) connected by the
170: wavy line. By introducing renormalized electron propagators and renormalized
171: hopping amplitudes each one of these groups gives rise to a diagram like the
172: one depicted in Fig. 1c. The remaining part of the calculation relies in
173: obtaining the expression of these diagrams in terms of Keldysh Green
174: functions.
175:
176: A further simplification of the calculation is obtained by considering the
177: wide band limit, i.e. to assume that the electron band width is much larger
178: than all other relevant energy scales involved in the problem. Within this
179: approximation the propagators for the isolated leads are given by
180:
181: \begin{eqnarray}
182: \hat{g}_{L,R}(\omega) &=& \left(
183: \begin{array}{cc}
184: g^{++}_{L,R}(\omega) & g^{+-}_{L,R}(\omega) \\
185: g^{-+}_{L,R}(\omega) & g^{--}_{L,R}(\omega)
186: \end{array}
187: \right) = \nonumber \\
188: & &\frac{i}{W} \left(
189: \begin{array}{cc}
190: 2 f_{L,R}(\omega) -1 & 2 f_{L,R}(\omega) \\
191: 2(f_{L,R}(\omega)-1) & 2f_{L,R}(\omega) - 1
192: \end{array}
193: \right) ,
194: \end{eqnarray}
195: where $f_{L,R}(\omega)$ are the Fermi distribution functions on the left and
196: right lead respectively.
197:
198: Another basic ingredient in the calculation is the convolution of electron
199: propagators with phase correlations $\delta g^{\alpha,\beta}_{L,R}(\omega) =
200: \int d\omega^{\prime} J^{\alpha,\beta}(\omega^{\prime})
201: g^{\alpha,\beta}_{L,R}(\omega+\omega^{\prime})$. In the wide band limit
202: these can be approximated as
203:
204: \begin{equation}
205: \delta \hat{g}_{L,R}(\omega) = \frac{i}{W} \left(
206: \begin{array}{cc}
207: f_{L,R}^+(\omega) + f_{L,R}^-(\omega) & 2 f_{L,R}^+(\omega) \\
208: 2 f_{L,R}^- & f_{L,R}^+(\omega) + f_{L,R}^-(\omega)
209: \end{array}
210: \right)
211: \end{equation}
212: where $f_{L,R}^{\pm}(\omega) = \int d\omega^{\prime} J(\omega^{\prime})
213: f_{L,R}(\omega\pm\omega^{\prime})$, $J(\omega)=J^{+-}(\omega)$ being the
214: Fourier transform of the phase correlation function.
215:
216: For an energy independent transmission coefficient one then obtains the
217: following expression for the correction to the current induced by the
218: environment:
219:
220: \begin{eqnarray}
221: \delta I(V) &=& \frac{e}{h} \tau (1 - \tau) \int d\omega \left[f_R(\omega)
222: \left(f^-_L(\omega) - f^+_L(\omega)\right) \right. \nonumber \\
223: && \left. - f_L(\omega) \left(f^-_R(\omega)
224: - f^+_R(\omega)\right) \right] \nonumber \\
225: && + \frac{e}{h} \tau^2 \int d\omega \left[f_L(\omega) \left(f^-_L(\omega)
226: - f^+_L(\omega)\right) \right. \nonumber \\
227: && \left. - f_R(\omega) \left(f^-_R(\omega) -
228: f^+_R(\omega)\right) \right]. \label{final-expression}
229: \end{eqnarray}
230:
231: By analyzing the Fermi factors, this expression can be decomposed as
232: $\delta I = \delta I^{\rightarrow}
233: - \delta I^{\leftarrow}$, where $\delta I^{\rightarrow}$ and
234: $\delta I^{\leftarrow}$
235: correspond to currents flowing in the two opposite directions.
236: Both terms can be interpreted as arising from
237: the coupling between the
238: contact current fluctuations and the phase fluctuations due to the
239: finite impedance of the environment. In fact, as described below,
240: this expression can be directly related to the current fluctuations.
241: In the absence of environment the noise spectrum
242: in a single channel contact is given by \cite{Khlus}
243:
244: \begin{eqnarray}
245: S(V,\Omega) &=& \frac{2e^2}{h} \tau (1 - \tau) \int d\omega \left[f_R(\omega)
246: \left(1 - f_L(\omega+\Omega)\right) \right. \nonumber \\
247: && \left. + f_L(\omega) \left(1 - f_R(\omega+\Omega)\right) \right] \nonumber \\
248: && + \frac{2e^2}{h} \tau^2 \int d\omega \left[f_L(\omega) \left(1 -
249: f_L(\omega+\Omega)\right) \right. \nonumber\\
250: && \left. + f_R(\omega) \left(1 -
251: f_R(\omega+\Omega)\right) \right] \nonumber \\
252: && + \left[\Omega \rightarrow -\Omega \right]. \label{noise-spectrum}
253: \end{eqnarray}
254:
255: On the other hand $S(V,\Omega) = \int dt e^{i\Omega t}
256: \left[K(t)+K(-t)\right]$ where
257: $K(t) = <\hat{I}(t)\hat{I}(0)> - <\hat{I}^2>$ is the current
258: correlation function. By comparing Eqs. (\ref{final-expression}) and
259: (\ref{noise-spectrum}) we arrive to the simple relation
260:
261: \begin{eqnarray}
262: e \left(\delta I^{\rightarrow} + \delta I^{\leftarrow} \right)
263: &=& \int dt J(t) \left[K(t) - K(-t) \right]. \label{general}
264: \end{eqnarray}
265:
266: This expression can be considered as a generalization of the
267: fluctuation-dissipation theorem to the present non-equilibrium situation. As
268: in the low impedance regime the coupling between the contact and its
269: environment is of the form $\hat{I}\hat{\phi}$ we expect this result to be
270: valid for a generic situation with the same type of
271: system-environment interaction. In particular the result (\ref{general})
272: would apply for any mesoscopic conductor that can be modeled as a collection
273: of channels.
274:
275: In order to understand the effects on the $I-V$ characteristic it is
276: instructive to consider first the case of an environment with just
277: a single mode at zero temperature, for which $J(\omega )=\pi
278: G_{0}Z_{0}(\delta (\omega -\omega _{0})-\delta (\omega ))/2$ \cite{ingold}.
279: For this model the conductance exhibits a discontinuity at
280: $eV=\hbar\omega _{0}$.
281: For $eV<\hbar\omega _{0}$
282: there is a reduction in the conductance $\delta G=-\pi
283: G_{0}^{2}Z_{0}\tau (1-\tau )/2$ while for $eV>\hbar\omega _{0}$, $\delta G=0$.
284: This simple case shows that the blockade is proportional to shot noise
285: and vanishes at perfect transmission.
286:
287: In the general situation the environment is characterized by a complex
288: impedance $Z(\omega )$. The phase correlation function is related to the
289: impedance by the expression \cite{ingold}
290:
291: \begin{equation}
292: J(t) = G_0 \int d\omega \frac{\mbox{Re} Z(\omega)}{\omega} \frac{e^{i\omega
293: t}-1}{1 - e^{-\beta \hbar \omega}} ,
294: \end{equation}
295: which at zero temperature leads to a correction in the differential
296: conductance given by
297:
298: \begin{equation}
299: \frac{\delta G}{G} = - G_0 (1 - \tau)
300: \int_{eV}^{\infty} d\omega \frac{\mbox{Re}%
301: Z(\omega)}{\omega} . \label{ohmic}
302: \end{equation}
303:
304: This is the same result one obtains for a tunnel junction except for
305: the reduction factor $(1 - \tau)$.
306: In the simple but realistic case in which the impedance $Z(\omega )$
307: is composed by the resistance $R$ of the leads embedding the contact in
308: parallel with the capacitance $C$ of the contact itself,
309: $Z(\omega)=R/(1+i\omega RC)$,
310: and the integral in (\ref{ohmic}) yields
311:
312: \begin{equation}
313: \frac{\delta G}{G} = - G_0 R (1-\tau) \ln\sqrt{1 +
314: \left(\frac{\hbar\omega_R}{eV} \right)^2 } ,
315: \label{ohmic2}
316: \end{equation}
317: where $\omega_R = 1/RC$. For finite temperature $\delta G$ can be evaluated
318: numerically from Eq. (\ref{final-expression}). Fig. 2 shows the evolution of
319: the correction to the differential conductance with temperature. For an
320: energy independent transmission the temperature dependence is the same as
321: for a tunnel junction except for a a global factor $(1-\tau)$. For
322: increasing temperatures the dip in the conductance at zero bias is
323: progressively washed out.
324:
325: The multichannel extension of these results is straightforward. One
326: can for instance analyze the case of a diffusive conductor by
327: replacing the reduction factor $(1-\tau)$ by its average over many
328: channels (which in the diffusive case is known to be $1/3$) and $R$ by the
329: resistance of the conductor itself \cite{zaikin}.
330: This analysis leads to quantitative
331: agreement with recent experimental results \cite{weber}.
332:
333: As a final remark we would like to stress that the results of the present
334: theory can be thoroughly tested experimentally using atomic contacts that
335: can be produced by scanning tunneling microscope or break junction techniques
336: \cite{ruitenbeek}.
337: These contacts accommodate a small number of channels,
338: and the ensemble of the transmissions $\{\tau _{i}\}$ can be
339: determined experimentally and varied over a wide range including
340: the $\tau \rightarrow 1$ limit \cite{scheer}.
341: Moreover, the impedance of
342: the environment embedding such contacts can be tuned within a desired range
343: using nanolithography \cite{goffman}. Experimental work along these lines is
344: currently under progress.
345:
346: In conclusion we have presented a theoretical analysis of the environmental
347: Coulomb blockade in coherent nanostructures. We have considered the
348: weak blockade regime and showed that the corrections in the current-voltage
349: characteristic can be related to the structure current fluctuations.
350: The blockade vanishes at perfect transmission as $\tau (1-\tau )$, in the
351: same way as shot noise. The temperature dependence of the conductance
352: is similar to the one observed in ultrasmall tunnel junctions. The present
353: calculation provides a first step towards the understanding of Coulomb
354: blockade effects in coherent nanostructures. Although derived for a
355: particular model it has been argued that the expression (\ref{general}) is
356: more general and valid for a generic mesoscopic conductor coupled to an
357: arbitrary external impedance. The strong blockade limit could be
358: addressed through the analysis of multiple photon processes along the
359: lines suggested in this work.
360:
361:
362: \acknowledgements
363: We acknowledge fruitful discussions with R. Cron, J.C. Cuevas, M.
364: Devoret, M. Goffman and P. Joyez. We also thank A.D. Zaikin for calling
365: to our attention the experimental results of Ref. \cite{weber}.
366: This work was partially supported by BNM and MAE through PICASSO and
367: the Spanish CICyT under contract PB97-0044.
368:
369: \begin{references}
370: \bibitem{noise1} For a review see Ya.M. Blanter and M. B\"uttiker,
371: Phys. Rep. {\bf 336}, 1 (2000).
372:
373: \bibitem{exp-noise} M.I. Reznikov et al. Phys. Rev. Lett. {\bf 75},
374: 3340 (1995); H.E. van den Brom and J.M. van
375: Ruitenbeek, Phys. Rev. Lett. {\bf 82}, 1526 (1999); R. Cron et al.,
376: submitted to Phys. Rev. Lett.
377:
378: \bibitem{devoret} M.H. Devoret et al., Phys. Rev. Lett. {\bf 64}, 1824
379: (1990); S.M. Girvin et al., Phys. Rev. Lett. {\bf 64}, 3183 (1990).
380:
381: \bibitem{junctions} A.N. Cleland, J.M. Schmidt and J. Clarke, Phys.
382: Rev. Lett. 64, 1565 (1990); T. Holst, et al., Phys. Rev. Lett. {\bf 73},
383: 3455 (1994).
384:
385: \bibitem{cuevas} J.C. Cuevas, A. Mart\'{\i}n-Rodero and A. Levy Yeyati,
386: Phys. Rev. B {\bf 54}, 7366 (1996).
387:
388: \bibitem{keldysh} L.V. Keldysh, Sov. Phys. JETP {\bf 20}, 1018 (1965).
389:
390: \bibitem{comment} It should be noticed that hopping events in the opposite
391: direction are connected by the correlator $P^{\alpha,\beta}$ while hopping
392: events in the same direction are connected by its inverse $%
393: 1/P^{\alpha,\beta} $.
394:
395: \bibitem{Khlus} V.A. Khlus, Sov. Phys. JETP {\bf 66}, 1243 (1987).
396:
397: \bibitem{ingold} G.-L. Ingold and Yu.V. Nazarov in {\it Single Charge
398: Tunneling}, edited by H. Grabert and M.N. Devoret (Plenum Press, New York,
399: 1992).
400:
401: \bibitem{ruitenbeek} J.M. van Ruitenbeek in {\it Mesoscopic electron
402: transport} Eds. L.L. Sohn, L.P. Kouwenhoven and G. Sh\"on, Kluwer Academic,
403: Dodrecht (1997).
404:
405: \bibitem{scheer} E. Scheer et al., Phys. Rev. Lett. {\bf 78}, 3535 (1997) and
406: E. Scheer et al., Nature {\bf 394}, 154 (1998).
407:
408: \bibitem{goffman} M. Goffman et al., Phys Rev. Lett. {\bf 85}, 170 (2000).
409:
410: \bibitem{zaikin} Coulomb blockade in a coherent conductor due to its own
411: resistance has been recently addressed by D.S. Golubev and A.D. Zaikin,
412: cond-mat/0010493. For this situation both theories yield similar
413: results.
414:
415: \bibitem{weber} H.B. Weber et al., cond-mat/0007033.
416:
417: \end{references}
418:
419: \begin{figure}[tbp]
420: \begin{center}
421: \leavevmode
422: \epsfysize=10cm
423: \epsfbox{rainbow.ps}
424: \end{center}
425: \caption{Diagrammatic expansion of the current in the Keldysh formalism: a)
426: Unlabeled lowest order diagrams in $\hat{H}_T$. Full lines with an arrow
427: indicate electron propagators, crosses indicate hopping events and wavy
428: lines correspond to environment correlators. b) Single photon processes up
429: to second order in $\hat{H}_T$. Dashed lines indicate phase correlation
430: functions. c) Renormalized diagrams arising from the addition of single
431: photon processes up to infinite order in the hopping. Double full lines with
432: an arrow indicate dressed electron propagators and shaded circles indicate
433: dressed hopping amplitudes.}
434: \end{figure}
435:
436: \begin{figure}[tbp]
437: \begin{center}
438: \leavevmode
439: \epsfysize=5cm
440: \epsfbox{ohmic.ps}
441: \end{center}
442: \caption{Correction to the conductance of a single conduction channel
443: with transmission $\tau$ due to an ohmic environment
444: for different temperatures $k_BT=$ 0 (full line), 0.005 (dotted line), 0.01
445: (short dashes) and 0.02 (long dashes) in units of $\hbar \protect\omega_R$.}
446: \end{figure}
447:
448: \end{document}
449: