1:
2: \documentclass[twocolumn,showpacs]{revtex4}
3:
4: \usepackage{graphicx}
5: \usepackage{dcolumn}
6: \usepackage{amsmath}
7:
8: \begin{document}
9:
10:
11: \title[Short Title]{Kondo Effect in Multiple-Dot Systems}
12:
13: \author{Rui Sakano}
14: \author{Norio Kawakami}
15: \affiliation{
16: Department of Applied Physics,
17: Osaka University, Suita, Osaka 565-0871, Japan\\
18: }
19: \date{\today}
20: \pacs{
21: 68.65.Hb, %Quantum dots
22: 71.27.+a, %Strongly correlated electron systems; heavy fermions
23: 72.15.Qm, %Scattering mechanisms and Kondo effect
24: 75.20.Hr %Local moment in compounds and alloys; Kondo effect, valence fluctuations, heavy fermions
25: }
26:
27: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
28: \begin{abstract}
29: We study the Kondo effect in multiple-dot systems for which
30: the inter- as well as intra-dot Coulomb repulsions are strong,
31: and the inter-dot tunneling is small.
32: The application of the Ward-Takahashi
33: identity to the inter-dot dynamical susceptibility
34: enables us to analytically calculate the conductance for a
35: double-dot system by using the Bethe-ansatz exact solution of the
36: SU(4) impurity Anderson model. It is clarified
37: how the inter-dot Kondo effect enhances or suppresses
38: the conductance under the control of
39: the gate voltage and the magnetic field. We
40: then extend our analysis to multiple-dot systems including
41: more than two dots, and discuss their characteristic transport
42: properties by taking a triple-dot system as an example.
43: \end{abstract}
44:
45: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
46: \maketitle
47:
48: %%%%%%%%%%%%%%%%%%%%%%%
49: \section{Introduction}
50: %%%%%%%%%%%%%%%%%%%%%%%
51:
52: Recent advances in semiconductor processing have made it possible
53: to fabricate various nanoscale materials with tunable
54: quantum parameters, revealing various aspects
55: of quantum mechanics.
56: Quantum dot
57: \cite{Tarucha,Reimann}
58: is one of the interesting nanoscale materials.
59: In particular, a lot of works on the Kondo effect in single
60: quantum dot systems have been done both theoretically and
61: experimentally
62: \cite{TK,Glazman,Kawabata,Izumida,Hofstetter,Eto,Oguri,DGG,Sasaki,Cho}.
63: More recently, double-dot systems or systems with more than
64: two dots have been
65: investigated
66: \cite{Nagaraja,WGW,Saraga}.
67: In this connection, the Kondo effect in
68: double-dot systems have been studied intensively
69: \cite{Aono,Georges,Jeong,Taka,Tanaka,Martin}.
70:
71: Most of the previous studies on multiple-dot systems have treated the
72: intra-dot Coulomb repulsion but have ignored the Coulomb repulsion
73: between quantum dots (inter-dot Coulomb repulsion). We especially focus
74: on the effect of the inter-dot Coulomb repulsion here
75: \cite{Wilhelm,Sun,Borda}
76: and study how such electron correlations affect transport properties.
77: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
78: Recently, Borda \textit{et al.} have investigated properties of the Kondo
79: effect in such double-dot systems with a magnetic field by
80: the numerical renormalization group method
81: \cite{Borda}, which may explain the Kondo effect observed experimentally
82: by Wilhelm \textit{et al.} in a double-dot system\cite{Wilhelm}.
83: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
84: A remarkable point in the above double-dot systems with
85: inter-dot Coulomb
86: repulsion is that enhanced charge fluctuations between the quantum dots
87: induce the ``inter-dot Kondo effect", which
88: plays an important role to determine transport properties of
89: the systems. Since the inter-dot Kondo effect is caused
90: not by spin fluctuations but by charge fluctuations
91: between two dots, its influence appears
92: significantly when the dots are connected in series.
93: In particular, by changing the gate voltage or
94: the magnetic field, one can control the conductance
95: via the inter-dot Kondo effect.
96:
97: In this paper, we investigate transport properties of
98: the double-dot systems with strong intra- and
99: inter-dot Coulomb repulsions mentioned above. We exploit a novel
100: method to treat the Kondo effect at absolute zero: the
101: application of the Ward-Takahashi identity enables us to
102: use the Bethe-ansatz exact solution of the
103: SU(4) impurity Anderson model to our double-dot system.
104: Our calculation clearly shows that the inter-dot Kondo
105: effect plays an important role on transport, which can be
106: controlled by the gate voltage and the magnetic field.
107: Our method is also applicable to multiple-dot systems including
108: more than two dots. We explore the
109: Kondo effect in such dot systems by taking a triple-dot
110: system as an example.
111:
112:
113: This paper is organized as follows.
114: In section \ref{sc:model}, we introduce the model Hamiltonian and
115: outline the method to treat our double-dot system:
116: how the Bethe-ansatz exact solution can be used to
117: compute the conductance at absolute zero. In section
118: \ref{sc:doubl-dotresult} we
119: discuss the results for the conductance
120: with particular emphasis on the gate-voltage
121: control and the magnetic-field control.
122: In section \ref{sc:maltiple} we extend our method to a
123: triple-dot system, and discuss its transport properties
124: on the basis of the exact solution. We also
125: mention
126: how we can treat generalized multiple-dot systems including
127: more dots. A brief summary is given in
128: section \ref{summary}.
129:
130: %%%%%%%%%%%%%%%%%%%%%%%%%%
131: %%%%%%%%%%%%%%%%%%%%%%%%%%
132: \section{Model and Method}\label{sc:model}
133: %%%%%%%%%%%%%%%%%%%%%%%%%%
134: %%%%%%%%%%%%%%%%%%%%%%%%%%
135:
136: We describe our model and method by taking
137: a double-dot system connected in series, which was
138: proposed by Borda \textit{et al.} \cite{Borda}.
139: The setup is schematically shown in FIG. \ref{fig:model},
140: %%%%%%%%%%%%%%%%%%%%%%%%%%%
141: \begin{figure}[bp]
142: \includegraphics[width=5.0cm]{DDregime}
143: \caption{Schematics of our double-dot system: two dots are connected via tunneling $t$, and each dot is connected to a lead via $V$.}
144: \label{fig:model}
145: \end{figure}
146: %%%%%%%%%%%%%%%%%%%%%%%%%%
147: where not only the ordinary Coulomb repulsion $U$, which works inside each
148: dot, but also $U'$ between the dots are introduced. We assume that
149: the inter-dot tunneling $t$ is small and the gate voltages are
150: such that the lowest-lying charged states are restricted
151: to the configurations of singly-occupied states, $(n_R,n_L)=(1,0)$
152: and $(0,1)$, where $n_{R(L)}$ is the number of extra electrons on the
153: right (left) dot. This situation is realized in the condition,
154: %%%%%%%%%%%%%%%%%%%%%%%%%%
155: \begin{equation}
156: |E(1,0)|, |E(0,1)| \ll U, U'
157: \end{equation}
158: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
159: where $E(n_R,n_L)$ is the energy level in the dots
160: measured from the common
161: chemical potential of the two leads. The
162: states $(1,0)$ and $(0,1)$ have a spin $S=1/2$, associated with
163: the extra electron on the double dots. Then at energies below the
164: charging energy of the double dots, dynamics of the double dots
165: is restricted to the subspace with the
166: 4 possible configurations of $\{ S_z=\pm 1/2 ; \ n_R-n_L = \pm 1 \}$
167: in addition to the unoccupied state of $n_R=n_L=0$.
168: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
169:
170: The above double-dot system, in which
171: both of intra- and inter-dot Coulomb repulsions are
172: sufficiently strong, may be modeled by
173: the highly correlated degenerate Anderson Hamiltonian
174: $H_A$ ($U,U'\to\infty$) supplemented by a inter-dot tunneling
175: term $H_T$,
176: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
177: \begin{eqnarray}
178: H_A&=& \sum_{\sigma,\tau} \int dx\, c_{\sigma \tau}^{\dagger}(x)
179: \frac{1}{i}\frac{\partial}{\partial x}c_{\sigma \tau}(x)
180: + \sum_{\sigma,\tau}\varepsilon_{d\tau}^{\sigma}
181: |\sigma\tau \rangle \langle \sigma \tau | \nonumber \\
182: &+& V\sum_{\sigma\tau} \int dx\, \delta(x)
183: [ | \sigma\tau \rangle \langle 0|c_{\sigma \tau}(x) + h.c.] , \\
184: \label{anderson}
185: %%%%%%%%%%%%%%%%%%%
186: H_T&= t & \sum_{\sigma,\tau\ne\tau'} | \sigma\tau \rangle \langle
187: \sigma \tau' |
188: %%%%%%%%%%%%%%%%%%%
189: \label{tunnel}
190: \end{eqnarray}
191: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
192: where $c_{\sigma \tau}^{\dagger}(x)(c_{\sigma \tau}(x))$ creates
193: (annihilates) a conduction electron at a position
194: $x$ with spin $\sigma$$(=\pm1/2)$ and
195: ``orbital index" $\tau$. Here we have represented
196: conduction electrons in the leads in the low-energy continuum limit by
197: assuming that its density of states is constant, $1/2\pi$. Also we have
198: introduced the orbital index $\tau=1/2$ (-1/2) to specify
199: an electron occupying the left (right) lead, which is
200: also used to label the left (right) dot.
201: A state $| \sigma\tau \rangle $ in the double dots
202: located at $x=0$ denotes a singly occupied
203: state and $|0\rangle$ denotes an unoccupied state.
204:
205: We will discuss transport properties of the system under the
206: gate-voltage control or the magnetic-field control.
207: It is thus convenient to
208: write down each energy level $\varepsilon_{d\tau}^{\sigma}$ as,
209: %%%%%%%%%%%%%%
210: \begin{eqnarray}
211: \varepsilon_{d\tau}^{\sigma} = \varepsilon + \delta E \cdot \tau + E_Z \cdot \sigma,
212: \end{eqnarray}
213: %%%%%%%%%%%%%
214: where $\delta E$ ($E_Z$) is the energy difference between
215: the two dots (Zeeman
216: energy). Note that the system possesses SU(4) symmetry
217: with respect to spin and orbital degrees of freedom
218: at $\delta E =E_Z =0$.
219:
220: We note here that
221: the Bethe-ansatz exact solution can be obtained for
222: the above four-component Anderson Hamiltonian $H_A$ \cite{schkawa},
223: which is referred to as the SU(4) Anderson model henceforth.
224: However, this method allows us
225: to calculate only static quantities, so that we cannot
226: apply the exact solution to transport quantities straightforwardly.
227: In the following, we outline
228: how we can overcome this difficulty to calculate the conductance.
229:
230: Let us begin with the expression for the conductance
231: in the above double-dot system connected in series, which is
232: obtained in the second order in the tunneling Hamiltonian $H_T$
233: between two dots,
234: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
235: \begin{equation}
236: G=-\frac{2\pi e^2}{h}t^2 \lim_{\omega \to 0}
237: \frac{\mbox{Im} \chi_{ops}(\omega)}{\omega}, \label{eq:conductance}
238: \end{equation}
239: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
240: where $\chi_{ops}(\omega)$ is the analytic continuation
241: ($i\omega_n \rightarrow \omega + i0$) of
242: the dynamical ``orbital pseudo-spin" susceptibility for the
243: SU(4) Anderson Hamiltonian $H_A$ (without $H_T$),
244: %%%%%%%%%%%%%%
245: \begin{eqnarray}
246: \chi_{ops}(i\omega_n) = \int_{0}^{\beta}d\tau e^{i\omega_n}
247: <{\cal T} \hat{T}_+(\tau)\hat{T}_-(0)>,
248: \end{eqnarray}
249: %%%%%%%%%%%%%%%%%
250: with the time-ordering operator ${\cal T}$.
251: The corresponding SU(2) operators are defined as
252: %%%%%%%%%%%%%%%%%%%%%%%
253: \begin{eqnarray}
254: \hat{T}_z & \equiv& \frac{1}{2}(\hat{n}_R-\hat{n}_L),\nonumber\\
255: \hat{T}_{\pm} & \equiv &\sum_{\sigma} |\sigma \pm1/2 \rangle \langle
256: \sigma \mp1/2 |.
257: \end{eqnarray}
258: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
259: These orbital pseudo-spin operators properly
260: describe inter-dot charge fluctuations.
261: As defined above, the eigenvalue $\tau=\pm 1/2$ of $T_z$
262: specifies which dot an electron occupies.
263: Eq.(\ref{eq:conductance}) means that the low-frequency inter-dot
264: ``orbital" susceptibility is essential
265: to determine the conductance.
266:
267: Although the low-frequency susceptibility is difficult to calculate in general,
268: we can make use of sophisticated techniques developed in the study of
269: the NMR relaxation rate in dilute magnetic alloys \cite{shiba,nakamura}:
270: the exact Ward-Takahashi relation for the low-frequency dynamical
271: pseudo-spin susceptibility is obtained, at zero temperature, as
272: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
273: \begin{eqnarray}
274: &&\lim_{\omega \to 0} \frac{\mbox{Im}\chi_{ops}
275: (\omega)}{\pi\omega} =
276: -\sum_{\sigma}\sum_{\tau, \tau'}(T_+)_{\tau\tau'}^2
277: K_{\tau \tau'}^\sigma,
278: \label{dynamical}
279: \end{eqnarray}
280: %%%%%%%%%%%%
281: with
282: %%%%%%%%%%%
283: \begin{eqnarray}
284: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
285: K_{\tau \tau'}^\sigma=
286: \left\{
287: \begin{array}{rl}
288: \rho_{d\tau}^{\sigma} \, \rho_{d\tau'}^{\sigma}
289: \left[ 1+ \frac{\Sigma_{d \tau}^{\sigma}
290: - \Sigma_{d \tau'}^{\sigma}}{\varepsilon_{d \tau}^{\sigma}
291: - \varepsilon_{d \tau'}^{\sigma}} \right]^2 \, \,
292: (\mbox{for}\ \varepsilon_{d \tau}^{\sigma} \neq
293: \varepsilon_{d \tau'}^{\sigma}) \\
294: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
295: \left[ \chi_{ops}^\sigma (0)\right]^2 \,\,
296: (\mbox{for}\ \varepsilon_{d \tau}^{\sigma} =
297: \varepsilon_{d \tau'}^{\sigma})
298: \end{array} \right. \label{eq:korringa}
299: \end{eqnarray}
300: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
301: where
302: $\rho_{d\tau}^{\sigma}$ ($ \Sigma_{d\tau}^{\sigma}$)
303: is the density of states (self energy) for an electron
304: at the Fermi level in the dot $\tau$.
305:
306: We note that the second line of Eq.(\ref{eq:korringa}) is
307: the well-kwon Korringa relation \cite{shiba} in the context of
308: NMR relaxation theory, and the first line is its extension to the
309: case having a finite energy-level splitting \cite{nakamura}.
310: Since the static susceptibility can be calculated by the
311: exact solution, we need to evaluate the density of states
312: $\rho_{d\tau}^{\sigma}$ and the self energy $ \Sigma_{d\tau}^{\sigma}$.
313: Fortunately, this can be done by
314: exploiting the Friedel sum rule. First recall that
315: the phase shift $\delta_{\tau}^{\sigma}$ of an
316: electron in the double dots at the Fermi level is
317: obtained from the
318: average number of electrons $\langle n_{d\tau}^{\sigma} \rangle$ in the
319: double dots: $\delta_{\tau}^{\sigma} = \pi \langle n_{d\tau}^{\sigma}\rangle$
320: (Friedel sum rule). Then, the density of states and the self energy
321: at the Fermi level is given by
322: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
323: \begin{eqnarray}
324: \rho_{d \tau}^{\sigma} &=& \frac{\sin^2 \delta_{\tau}^{\sigma}}{\pi \Delta} ,
325: \label{eq:fermil1} \\
326: \Sigma_{d \tau}^{\sigma} &=& \Delta \cot
327: \delta_{\tau}^{\sigma} - \varepsilon_{d \tau}^{\sigma} \label{eq:fermil2}
328: \end{eqnarray}
329: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
330: with the resonance width
331: $\Delta$ due to the mixing $V$.
332: %%%%%%%%%%%%%%%%%%%%%%%%%%%
333: Note that the electron number $\langle n_{d\tau}^{\sigma} \rangle$
334: can be evaluated by the exact solution.
335:
336: Combining all the above relations, we can compute
337: the conductance at zero temperature:
338: the quantities in the right hand side of Eq.(\ref{eq:korringa}),
339: $\rho_{d\tau}^{\sigma}$, $\Sigma_{d \tau}^{\sigma}$,
340: $\chi^{\sigma}_{ops}(0)$, can be evaluated by means of
341: the Bethe-ansatz solution of the SU(4) Anderson
342: model \cite{schkawa}.
343:
344: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
345: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
346: \section{double-dot system}\label{sc:doubl-dotresult}
347: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
348: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
349:
350: We study the conductance in several cases in our double-dot
351: system, which
352: are schematically shown in FIG. \ref{fig:energystates}.
353: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
354: \begin{figure}
355: \includegraphics[width=8.0cm]{schemetics}
356: \caption{Schematic description of the energy states in
357: the double dots connected in series: (a) SU(4) symmetric
358: case, (b) asymmetric
359: case, (c) symmetric case in a magnetic field, (d) symmetric
360: case in a strong magnetic field, (e) asymmetric
361: case in a strong magnetic field.}
362: \label{fig:energystates}
363: \end{figure}
364: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
365:
366: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
367: \subsection{Charge fluctuations in symmetric double dots}
368: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
369:
370: Let us start with the double-dot system shown
371: in FIG. \ref{fig:energystates}(a), where the energy levels
372: of two dots are same, which we refer to as the
373: symmetric dots in this paper: there are
374: 4 degenerate electron states including spin degrees of freedom.
375: In this case, from the
376: expressions (\ref{eq:conductance}), (\ref{dynamical}) and (\ref{eq:korringa})
377: we write down the conductance
378: in the absence of the magnetic field as,
379: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
380: \begin{eqnarray}
381: G &=& 2\pi^2 \cdot \frac{e^2}{h}t^2 \sum_{\sigma}
382: \left[ \chi^{\sigma }_{ops}(0) \right]^2 \nonumber \\
383: &=& 4\pi^2 \cdot \frac{e^2}{h}t^2
384: \left[ \chi^{\uparrow }_{ops}(0) \right]^2 .
385: \end{eqnarray}
386: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
387: By computing static pseudo-spin susceptibility
388: $\chi^{\uparrow}_{ops}(0)$
389: by means of the Bethe-ansatz solution of the SU(4)
390: Anderson model, we evaluate
391: the conductance as a function of the effective
392: energy level $\varepsilon^*$ \cite{eflevel}.
393: The results are shown in FIG. \ref{fig:vf}.
394: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
395: \begin{figure}[bt]
396: \includegraphics[width=6.5cm]{vf-g}
397: \caption{(a) Conductance in the cases of the SU(4) and SU(2) symmetric
398: double-dot systems as a function of the renormalized energy
399: level $\varepsilon^*$.
400: (b) conductance on log scale: we can see distinct exponential
401: dependence between the SU(4) (zero field) and SU(2)
402: (strong field) cases in the Kondo regime.}
403: \label{fig:vf}
404: \end{figure}
405: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
406: When the dot-level $\varepsilon^*$ is above the Fermi level,
407: the conductance is small, since the resonant tunneling
408: does not occur. As $\varepsilon^*$ goes down through the Fermi
409: level, the conductance is enhanced by
410: the Kondo effect, which is analogous to an ordinary
411: single dot case. However, in contrast to the single dot system,
412: for which the conductance is saturated in the Kondo limit with deep
413: $\varepsilon^*$,
414: it continues to increase exponentially. The increase
415: is caused by the inter-dot charge fluctuations enhanced by the
416: inter-dot ``orbital" Kondo effect
417: \cite{Wilhelm,Sun,Borda}. Since the
418: static susceptibility $\chi_{ops}^\sigma(0)$
419: is inversely proportional to the Kondo temperature
420: $T_K \sim \exp(-\Delta/\varepsilon^*)$,
421: the conductance has the exponential dependence like
422: $\exp(-2\Delta/\varepsilon^*)$.
423:
424: Note that the ordinary spin Kondo effect and the
425: inter-dot Kondo effect both emerge in the above SU(4) symmetric
426: case. Therefore, in order to see the above
427: characteristic enhancement of the conductance more clearly,
428: we consider an extreme case with strong
429: magnetic fields, where the
430: spin Kondo effect is completely suppressed.
431: Shown in FIG. \ref{fig:vf}(b) is the conductance in strong
432: magnetic fields (corresponding to FIG. \ref{fig:energystates}(d)).
433: We can see the enhancement of the conductance due to
434: inter-dot Kondo effect with SU(2) symmetry. In this case, the corresponding
435: Kondo temperature is given by $T_K \sim \exp(-2\Delta/\varepsilon^*)$, so that
436: the increase of the conductance, $\sim \exp(4\Delta/\varepsilon^*)$,
437: is more significant in comparison
438: with the zero field case.
439: These results are indeed seen in log-scale plots
440: given in FIG. \ref{fig:vf}, which clearly features
441: the exponential dependence
442: of the conductance in the Kondo regime.
443:
444: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
445: \subsection{Symmetric double dots: magnetic-field control}
446: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
447:
448: It is seen from FIG. \ref{fig:vf} that in the Kondo regime with
449: deep dot levels, the conductance in the SU(2) case
450: (strong field) is larger than that in the SU(4) case (zero field).
451: This implies that the conductance may be monotonically enhanced
452: in the presence of a magnetic field. To clarify
453: this point, we focus on the field-dependence of the
454: conductance for the SU(4) symmetric double-dot system (shown
455: in FIG. \ref{fig:energystates}(c)) in the Kondo regime.
456: Following the way outlined above, we can derive the conductance
457: in this case,
458: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
459: \begin{eqnarray}
460: G = 2\pi^2 \cdot \frac{e^2}{h}t^2 \left[ \left\{ \chi^{\uparrow }_{ops}(0)
461: \right\}^2 + \left\{ \chi^{\downarrow }_{ops}(0) \right\}^2 \right].
462: \end{eqnarray}
463: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
464: By exploiting the exact solution of the SU(4) Anderson model in the
465: Kondo regime (so-called Coqblin-Schrieffer model), we compute
466: the conductance
467: as a function of the Zeeman splitting $E_Z$, which is shown
468: in FIG. \ref{fig:ddmg}. Also,
469: the effective Kondo temperature $T_K(E_Z)$ is plotted as
470: a function of the Zeeman splitting on the log-log scale
471: in FIG. \ref{fig:loglogkt}. Here we assume that the direction of the magnetic field is parallel to spin$\uparrow$. It is seen that the magnetic fields
472: enhance the conductance, in contrast to the ordinary
473: Kondo effect in a single-dot system.
474: The inter-dot Kondo effect is caused by the degenerate energy levels
475: in two dots, which still possess SU(2) symmetry in
476: strong magnetic fields. Since the pseudo-spin susceptibility
477: $\chi_{ops}^{\uparrow}(0)$
478: increases with the increase of the field, thus resulting in the
479: enhanced conductance. In strong fields,
480: the effective Kondo temperature, which is defined by
481: the inverse of $\chi_{ops}^{\sigma}(0)$, is given by \cite{schkawa},
482: \begin{eqnarray}
483: T_K(E_Z)/T_K(0) \sim (E_Z/T_K(0))^{-1},
484: \end{eqnarray}
485: %%%%%%%%%%%%%%
486: so that the conductance increases as
487: \begin{eqnarray}
488: G \sim (T_K(E_Z)/T_K(0))^{-2}\sim (E_Z/T_K(0))^{2}.
489: \end{eqnarray}
490: %%%%%%%%%%%%%%%%%%%%%
491: Here we note
492: that the conductance for electrons
493: with spin parallel (anti-parallel)
494: to the magnetic field increases (decreases).
495: This effect might be utilized for spin-current control
496: by using double-dot systems.
497: The above results agree with those of
498: Borda \textit{et al.} obtained by the
499: numerical renormalization group analysis \cite{Borda}.
500:
501: %%%%%%Feb 5th 2005
502: It is to be noted here that
503: the SU(4) Kondo resonance has been observed not only in a
504: double-dot system \cite{Wilhelm} but also
505: in a single vertical quantum dot whose symmetric shape
506: gives rise to SU(4) internal degrees of freedom
507: \cite{Sasaki2}. Also, STM experiments on a Cr(001) surface
508: have found the SU(4) Kondo resonance, where the degeneracy
509: of $d_{xz}$ and $d_{yz}$ states gives the additional orbital
510: degrees of freedom \cite{Zhuravlev}.
511: Our results are also consistent with these findings.
512: %%%%%%%
513: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
514: \begin{figure}[bthp]
515: \includegraphics[width=6.0cm]{ddmg}
516: \caption{Conductance as a function of the Zeeman splitting
517: in the Kondo regime.
518: We also show the contribution of the electrons with spins
519: parallel (anti-parallel) to the magnetic field.
520: Note that
521: $T_K=T_K(E_Z=0)$ is the Kondo
522: temperature of the SU(4) Anderson model.}
523: \label{fig:ddmg}
524: \end{figure}
525: %%%%%%%%%%%%%%%%%%%%%%%%%%%
526: \begin{figure}[bthp]
527: \includegraphics[width=5.5cm]{loglogkt}
528: \caption{The effective
529: Kondo temperature as a function of the Zeeman splitting on
530: log-log scale.}
531: \label{fig:loglogkt}
532: \end{figure}
533: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
534:
535:
536:
537: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
538: \subsection{Asymmetric double dots: gate-voltage control}
539: %%%%%%%%%%%%%%%%%%%%%%%%%%%
540:
541: Next, we consider how the conductance is influenced by
542: the energy-level difference between
543: the two dots, which is controlled by changing the gate voltage
544: of each dot. We study two typical cases in the Kondo regime:
545: zero magnetic field (FIG. \ref{fig:energystates}(b))
546: and strong magnetic fields (FIG. \ref{fig:energystates}(e)).
547:
548: From the expressions (\ref{eq:conductance})
549: and (\ref{dynamical})-(\ref{eq:fermil2}), the conductance at zero field
550: is written as
551: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
552: \begin{eqnarray}
553: G=4 \cdot \frac{e^2}{h}t^2\frac{\sin^2(\pi\langle n_{L}^{\uparrow}
554: \rangle - \pi\langle n_{R}^{ \uparrow } \rangle)}{\delta E^2}.
555: \end{eqnarray}
556: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
557: We compute the conductance as a function
558: of energy differences $\delta E$. We also study the conductance
559: in strong magnetic fields, where the system is completely polarized,
560: and the remaining inter-dot charge fluctuations are
561: described by the SU(2) Kondo model subjected to the energy
562: difference $\delta E$\cite{Wilhelm,Sun,Borda}.
563: %%%%%%%%%%%%%%%%%%%%%%%%%
564: \begin{figure}[btp]
565: \includegraphics[width=6.0cm]{dif}
566: \caption{Conductance as a function of the energy difference
567: $\delta E$ between two dots. We
568: take $T_K=T_K(E_Z=0)$ for the SU(4) double-dot case
569: (zero field)
570: and $T_K=T_K(E_Z=\infty)$ for the SU(2) double-dot case
571: (strong fields).}
572: \label{fig:dif}
573: \end{figure}
574: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
575: The results obtained in both cases are shown in FIG. \ref{fig:dif}.
576: In contrast to the magnetic-field dependence discussed above,
577: the conductance decreases monotonically as a function of
578: $\delta E$. The decrease is caused by the suppression of
579: the inter-dot Kondo effect in the presence
580: of finite energy difference $\delta E$.
581:
582: We note here that the ordinary ``spin" Kondo effect
583: still persists even in finite $\delta E$, giving rise to the
584: enhanced spin fluctuations. Although such enhancement in spin
585: fluctuations may not be
586: observed in transport properties, it should show up if we observe
587: the NMR relaxation rate in the double-dot system. For example,
588: the $\delta E$-dependence of the NMR relaxation rate is
589: exactly given by the function shown in FIG. \ref{fig:ddmg}:
590: for large $\delta E$, it is enhanced as $(\delta E)^2$.
591:
592:
593: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
594: \section{triple-dot system}\label{sc:maltiple}
595: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
596:
597: We now discuss how the above method can be used to calculate the
598: conductance for multiple-dot systems with more than two dots. Here,
599: we deal with a triple-dot system, and then briefly outline how to
600: extend the method to $N$-dot systems.
601: We will see that the conductance exhibits some characteristic
602: properties
603: under the control of the gate voltage and the magnetic field.
604:
605:
606: Let us consider a triple-dot system, for which three dots
607: and three leads as arranged as shown in FIG. \ref{fig:tripledot}.
608: Inter-dot tunneling $t$ (intra- and inter-dot Coulomb repulsions)
609: is assumed to be sufficiently small (large) here again. Therefore,
610: one of the three dots can accommodate an electron thanks to
611: strong intra- and inter-dot correlations.
612: We focus on the Kondo regime, where the energy levels
613: in the dots are sufficiently lower than the Fermi level.
614: We note that a similar but different triple-dot system has been
615: proposed recently \cite{kuzmenko} and
616: its symmetry properties have been discussed.
617:
618: In the second order in tunneling $t$, we can calculate
619: the conductance between two leads in the triple-dot system
620: by the exact solution of the SU(6) Anderson model, because
621: there are 6 available electron states including spin
622: degrees of freedom in the three dots.
623: In this case, we can still utilize the formula Eq.(\ref{eq:conductance})
624: to calculate the conductance,
625: where the strong inter-dot correlations among three dots are
626: incorporated via the inter-dot susceptibility $\chi_{ops}$
627: between two dots through which electric currents flow.
628: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
629:
630: \begin{figure}[btp]
631: \includegraphics[width=4.0cm]{tripledot}
632: \caption{Schematics of our triple-dot system: three dots are
633: connected via small tunneling $t$, and each dot is connected to a
634: lead via tunneling $V$. Inter- as well as intra-dot
635: Coulomb repulsions are assumed to be sufficiently strong.}
636: \label{fig:tripledot}
637: \end{figure}
638:
639: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
640: \subsection{gate-voltage control}
641: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
642:
643: Transport properties for the above three-dot systems
644: depend on how the current is observed.
645: To be specific, we change the gate voltage
646: attached to the dot 3 with keeping the voltage in the
647: dots 1 and 2 fixed, and observe the conductance between
648: the leads 1 and 2 as well as between the leads 1 and 3.
649: \\
650: \\
651: The computed conductance
652: is shown in FIG. \ref{graph:tricon}(a) as a function of
653: the energy difference $\delta E$ between the energy level in
654: the dot 3 and those in the dots 1 and 2.
655: We set the sign of $\delta E$ positive when
656: the energy level in the dot 3 is higher than
657: the others.
658:
659: Let us first observe the current between the leads 1 and 2.
660: It is seen that the conductance increases
661: with the increase of $\delta E \, (>0)$. This
662: increase is attributed to the enhancement of the inter-dot
663: Kondo effect in the presence of the energy deference, which is
664: similar to that for the
665: double dots in magnetic fields discussed in the previous section.
666: At $\delta E=0$, the current flows via an SU(6) Kondo resonance
667: (i.e. 6-fold degenerate Kondo resonance). On the other hand, for
668: large $\delta E$, the SU(4) Kondo effect is realized within
669: 4 lower states in the dots 1 and 2. This gives the enhancement
670: of the inter-dot susceptibility between the dots 1 and 2,
671: resulting in the increase of the conductance. According to
672: the exact solution of the SU(6) Anderson model \cite{sch2},
673: the effective Kondo temperature
674: for large $\delta E$ is given by,
675: %%%%%%%%%%%%%%%%%%%%%%
676: \begin{eqnarray}
677: T_K^{(eff)} (\delta E)/T_K^{(6)} \sim (\delta E/T_K^{(6)})^{-1/2},
678: \end{eqnarray}
679: %%%%%%%%%%%%%%%%%%%%%
680: and the corresponding conductance is
681: %%%%%%%%%%%%%%%%%%%%%%
682: \begin{eqnarray}
683: G \sim \delta E/T_K^{(6)},
684: \end{eqnarray}
685: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
686: where $T_K^{(6)}=T_K^{(eff)}(\delta E=0)$ is the Kondo temperature for SU(6) triple-dot systems.
687:
688: On the other hand, for
689: $\delta E <0$, the SU(2) spin Kondo effect occurs in the lower
690: 2 levels in the dot 3,
691: while the spin Kondo effects in the dots 1 and 2 are suppressed
692: because the number of electrons in the dots 1 and 2
693: decreases (see FIG. \ref{graph:tricon}(b)).
694: Also, the inter-dot Kondo effect
695: between the dots 1 and 2 is suppressed. As a result
696: the conductance decreases when the current is observed
697: between leads 1 and 2.
698:
699: If the current is observed between the leads 1 and 3, distinct
700: properties appear in the conductance.
701: As seen from FIG. \ref{graph:tricon}(a), for large $|\delta E|$
702: (irrespective of its sign),
703: the conductance decreases because the energy difference suppresses
704: the inter-dot Kondo effect between the dots 1 and 2. Notice that around
705: $\delta E/T_K^{(6)} \sim -1$, the conductance
706: has a maximum structure, where charge fluctuations between the
707: dots 1 and 3 are slightly enhanced. Anyway, the
708: conductance exhibits behavior similar to that observed in
709: the double-dot case under the gate-voltage control.
710:
711: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
712: \begin{figure}[btp]
713: \includegraphics[width=6.0cm]{tricond2}
714: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
715: \caption{(a)Conductance as a function of the energy difference. Here, $T_K^{(6)}$ is the
716: Kondo temperature
717: for SU(6) triple-dot systems. (b)the number of electrons for the asymmetric
718: triple-dot as a function of the energy difference. The solid line is
719: the number of electrons in the dot 1 or the dot 2 per spin and
720: the dashed line is that in the dot 3.}
721: \label{graph:tricon}
722: \end{figure}
723: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
724:
725: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
726: \subsection{magnetic-field control}
727: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
728:
729: Let us now discuss how a magnetic field affects transport properties.
730: For simplicity, we assume that the
731: energy levels of three dots are same (symmetric dots).
732: The computed conductance between two leads
733: under magnetic fields is shown
734: in FIG. \ref{graph:triconmag}.
735: The conductance increases
736: as the Zeeman splitting $E_Z$ increases although the Kondo effect due to spin fluctuations
737: are suppressed by the field. As discussed in the
738: previous section, this enhancement is caused by
739: the inter-dot Kondo effect among three dots. For large magnetic fields,
740: half of the internal degrees of freedom are quenched, so that
741: the symmetry of the system changes from SU(6) to SU(3).
742: As a result, the SU(3) Kondo effect caused by inter-dot charge
743: fluctuations
744: is enhanced, and therefore the conductance is increased.
745: The effective Kondo temperature in large fields is given as,
746: %%%%%%%%%%%%%%%%%%%%%%
747: \begin{eqnarray}
748: T_K^{(eff)}(E_Z) /T_K^{(6)} \sim (E_Z/T_K^{(6)})^{-1},
749: \end{eqnarray}
750: %%%%%%%%%%%%%%%%%%%%%
751: and thus the conductance is enhanced like
752: %%%%%%%%%%%%%%%%%%%%%%
753: \begin{eqnarray}
754: G \sim (E_Z/T_K^{(6)})^{2}.
755: \end{eqnarray}
756:
757: %%%%%%%%%%%%%%%%%%%%%%%%%%%
758: \begin{figure}[btp]
759: \includegraphics[width=5.0cm]{su63g2}
760: \caption{Conductance as a function of the Zeeman splitting.}
761: \label{graph:triconmag}
762: \end{figure}
763: %%%%%%%%%%%%%%%%%%%%%%%%%%%
764:
765: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
766: \subsection{generalization to systems with more dots}
767: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
768:
769: We can generalize our method to systems with more than three dots:
770: a lead is attached to each dot, where the electrons feel
771: strong intra- and inter-dot Coulomb repulsions.
772: All the dots are connected to each other
773: via small inter-dot coupling $t$.
774:
775: In similar manners mentioned above, we can calculate the conductance
776: in such multiple-dot systems. The calculation can be
777: done by using the formula Eq.(\ref{eq:conductance}) in the second
778: order in tunneling $t$, where all the correlation effects are
779: incorporated through the dynamical susceptibility.
780: We can use the exact solution of the SU($2N$) Anderson model
781: \cite{schkawa}
782: for an $N$-dot system.
783: The conductance shows similar properties to those
784: observed in the double and triple dots: if we change the gate voltage of the dot $\tau$, the conductance between the lead $\tau$ and one of the other leads is generally suppressed, while it is enhanced otherwise.
785:
786:
787: In this paper we have assumed small inter-dot
788: coupling $t$, and calculated the conductance up to
789: $t^2$. It should be mentioned that for a system with
790: more than two dots, a Fano-type interference effect may
791: emerge in higher order terms in $t$. This interference effect
792: may give another interesting aspect of multiple-dot
793: systems, which is to be studied in the future work.
794:
795:
796: %%%%%%%%%%%%%%%%%%%%%%%
797: \section{Summary}\label{summary}
798: %%%%%%%%%%%%%%%%%%%%%%
799:
800: We have studied transport properties in the double-dot system
801: connected in series that possesses not only intra- but also
802: inter-dot Coulomb repulsions. It has been shown that the
803: application of the Ward-Takahashi
804: identity enables us to use the exact solutions of the
805: Anderson model for calculations of the conductance at zero
806: temperature. We have clarified
807: how the inter-dot Kondo effect affects the conductance
808: under the gate-voltage control and the magnetic-field
809: control. In particular, the conductance is
810: decreased by the suppression of the
811: inter-dot Kondo effect in the gate-voltage control, whereas
812: it is increased by the enhanced Kondo effect in
813: the presence of magnetic fields. The latter conclusion is consistent
814: with the results of the numerical renormalization
815: group. The method has also been applied to calculate
816: the conductance in multiple-dot systems including more
817: than two dots. By taking a triple-dot
818: system as an example, we have shown how the conductance is controlled
819: by tuning the inter-dot Kondo effect.
820:
821: Naively, it seems not easy to observe the Kondo effect
822: in multiple-dot systems (more than two dots) experimentally.
823: We would like to mention, however, that the Kondo
824: temperature in multiple-dot systems
825: can be much higher than that in
826: single-dot systems when the inter-dot repulsion is relevant,
827: as assumed in this paper. Therefore, if such multiple-dot
828: system could be fabricated, the Kondo effect may be possibly
829: observed even in multiple-dot systems considered here.
830:
831: Finally a comment is in order on the ordinary ``spin" Kondo effect in
832: our multiple-dot system. We have focused on the inter-dot ``orbital"
833: Kondo effect in this paper, which directly affects transport properties.
834: Concerning the spin Kondo effect,
835: the impacts of the gate voltage
836: and the magnetic field appear differently from the
837: inter-dot Kondo effect, e.g. the magnetic field
838: (gate-voltage difference) suppresses (enhances)
839: the spin Kondo effect. If we use the dynamical spin
840: susceptibility instead of the pseudo-spin susceptibility, the present
841: analysis can be straightforwardly applied to low-frequency
842: dynamics such as
843: the NMR relaxation rate, which may be important to discuss
844: an application to quantum-bits in quantum
845: computation. In fact, the expression Eq.(\ref{dynamical})
846: gives the NMR relaxation rate for the double-dot system,
847: if the dynamical susceptibility is regarded as the spin
848: susceptibility. It is of particular interest that the NMR
849: relaxation rate in our multiple-dot systems can be controlled by
850: the gate voltage, e.g. the difference in the energy-levels
851: of double dots can enhance the relaxation rate.
852: \\
853: \\
854:
855:
856: After the completion of this paper, we became aware of a recent preprint which deals with the SU(4) Kondo effect in a slightly different model\cite{Chud}.
857:
858: %%%%%%%%%%%%%%%%%%%%%%%%%%
859: \begin{acknowledgments}
860: We would like to express our sincere thanks to M. Eto
861: for valuable discussions.
862: R.S. also thanks H. Akai for fruitful discussions and supports.
863: \end{acknowledgments}
864:
865: %%%%%%%%%%%%%%%%%%%%%%%%%
866: \begin{thebibliography}{99}
867: %%%%%%%%%%%%%single dot%%%%%%%%%%%%%%%%%%%%%%%%%
868:
869: \bibitem{Tarucha}
870: L. P. Kouwenhoven, D. G. Austing and S. Tarucha, Rep. Prog. Phys. \textbf{64}, 701 (2001)
871:
872: \bibitem{Reimann}
873: S. M. Reimann, Rev. Mod. Phys. \textbf{74}, 1283 (2002)
874:
875: %%%%%%%%%%%%%%%%%%%single dot Kondo%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
876: \bibitem{TK}
877: T. K. Ng and P. A. Lee, Phys. Rev. Lett. \textbf{61}, 1768 (1988)
878:
879: \bibitem{Glazman}
880: L.I. Glazman and M. \'E. Ra\u{i}kh, JETP Lett. \textbf{47}, 452 (1988)
881:
882: \bibitem{Kawabata}
883: A. Kawabata, J. Phys. Soc. Jpn. \textbf{60}, 3222 (1991)
884:
885: \bibitem{Izumida}
886: W. Izumida, O. Sakai and Y. Shimizu, J. Phys. Soc. Jpn. \textbf{67}, 2444 (1998)
887:
888: \bibitem{Hofstetter}
889: W. Hofstetter, J. K\"onig and H. Schoeller, Phys. Rev. Lett. \textbf{87}, 156803 (2001)
890:
891: \bibitem{Eto}
892: M. Eto and Y. V. Nazarov, Phys. Rev. B \textbf{64}, 085322 (2001)
893:
894: \bibitem{Oguri}
895: A. Oguri, J. Phys. Chemi. Solids \textbf{63}, 1591-1594 (2002); Phys. Rev. B \textbf{64}, 153305 (2001)
896:
897: \bibitem{DGG}
898: D. Goldhaber-Gordon, H. Shtrikman, D. Mahalu, D. Abusch-Magder, U. Meirav and M. A. Kastner, Nature (London) \textbf{391}, 156 (1998); S. M. Cronenwett, T. H. Oosterkamp, L. P. Kouwenhoven, Science \textbf{281}, 540 (1998); J.Schmid, J. Weis, K. Eberl and K. v. Klizing: Physuca B \textbf{256}, 182 (1998)
899:
900: \bibitem{Sasaki}
901: S. Sasaki, S. De Franceschi, J. M. Elzerman, W. G. van der Wiel, M. Eto, S. Tarucha and L. P. Kouwenhoven, Nature, \textbf{405}, 764 (2000)
902:
903: \bibitem{Cho}
904: S. Y. Cho, H. Q. Zhou and R. H. McKenzie, Phys. Rev. B \textbf{68}, 125327 (2003)
905:
906:
907: %%%%%%%%%%%%%%%%%%double and maltiple dot%%%%%%%%%%%%%%%%%%%%%%%%%%
908: \bibitem{Nagaraja}
909: S. Nagaraja, J. P. Leburton and R. M. Martin, Phys. Rev. B \textbf{60}, 8759 (1999)
910:
911:
912: \bibitem{WGW}
913: W. G. van der Wiel, S. De Franceschi, J. M. Elzerman, T. Fujisawa, S. Tarucha and L. P. Kouwenhoven, Rev. Mod. Phys. \textbf{75}, 1 (2003)
914:
915: \bibitem{Saraga}
916: Daniel S. Saraga and Daniel Loss, Phys. Rev. Lett. \textbf{90}, 166803 (2003)
917:
918: %%%%%%%%%%%%%%%%%%double-dot Kondo%%%%%%%%%%%%%%%%%%%%%%%%%%%%
919:
920: \bibitem{Aono}
921: T. Aono, M. Eto and K. Kawamura, J. Phys. Soc. Jpn. \textbf{67}, 1860 (1998)
922:
923: \bibitem{Georges}
924: A. Georges and Y. Meir, Phys. Rev. Lett. \textbf{82}, 3508 (1999)
925:
926: \bibitem{Jeong}
927: H. Joeng, A. M. Chang and M. R. Melloch, Sience \textbf{293}, 2221 (2001)
928:
929: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
930: \bibitem{Taka}
931: K. Kikoin and Y. Avishai, Phys. Rev. Lett. 86, 2090 (2001);
932: Phys. Rev. B 65, 115329 (2002);
933: Y. Takazawa, Y. Imai, and N. Kawakami,
934: J. Phys. Soc. Jpn. \textbf{71}, 2234 (2002).
935:
936: \bibitem{Tanaka}
937: Y. Tanaka and N. Kawakami,
938: J. Phys. Soc. Jpn. \textbf{73}, 2795 (2004).
939:
940: \bibitem{Martin}
941: G. B. Martins, C. A. B\"usser, K. A. Al-Hassanieh, A. Moreo and E. Dagotto, Phys. Rev. Lett. \textbf{94}, 026804 (2005)
942: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
943: \bibitem{Wilhelm}
944: U. Wilhelm and J. Weis, Physica E \textbf{6}, 668 (2000)
945:
946: \bibitem{Sun}
947: Q. F. Sun and H. Guo, Phys. Rev. B \textbf{66}, 155308 (2002)
948:
949: \bibitem{Borda}
950: L. Borda, G. Zarand, W. Hofstetter, B. I. Halperin, and J. von Delft, Phys. Rev. Lett. \textbf{90}, 26602 (2003)
951:
952: \bibitem{schkawa}
953: P. Schlottmann, Phys. Rev. Lett. \textbf{50}, 1697 (1983);
954: N. Kawakami and A. Okiji, J. Phys. Soc. Jpn. \textbf{54}, 685 (1985)
955:
956: \bibitem{shiba}
957: H. Shiba, Prog. Theor. Phys. \textbf{54} 967 (1975)
958:
959: \bibitem{nakamura}
960: A. Nakamura, N. Kawakami and A. Okiji, J. Phys. Soc. Jpn. \textbf{56}, 3667 (1987) ; R. Sakano and N. Kawakami, to be published in Proceedings of International Symposium on the Creation of Novel Nanomaterials, Journal of Electron Microscopy, Supplement (2005)
961:
962: \bibitem{eflevel}
963: F. D. M. Haldane, Phys. Rev. Lett. \textbf{40}, 416 (1978); J. Phys. C\textbf{11}, 5015 (1978)
964:
965: \bibitem{Sasaki2}
966: S. Sasaki, S. Amaha, N. Asakawa, M. Eto and S. Tarucha, Phys. Rev. Lett. \textbf{93}, 017205 (2004) ; M. Eto, J. Phys. Soc. Jpn. \textbf{74}, 95 (2004)
967:
968: \bibitem{Zhuravlev}
969: O. Yu. Kolesnychenko, R. de Kort, M. I. Katsnelson, A. I. Lichtenstein and H. Van Kempen, Nature (London) \textbf{415}, 507 (2002);
970: A. K. Zhuravlev, V. Yu. Irkhin, M. I. Katsnelson and A. I. Lichtenstein, Phys. Rev. Lett. \textbf{93}, 236403 (2004)
971:
972: \bibitem{kuzmenko}
973: T. Kuzmenko, K. Kikoin and Y. Abishai, cond-mat/0412527; see also, Phys. Rev. Lett. \textbf{89}, 156602 (2002)
974:
975: \bibitem{sch2}
976: P. Schlottmann, Phys. Rev. B \textbf{30}, 1454 (1984)
977:
978: \bibitem{Chud}
979: A. L. Chudnovskiy, preprint, cond-mat/0502282
980:
981: \end{thebibliography}
982: \end{document}
983:
984:
985:
986:
987: