1: \documentclass[12pt]{iopart} %>>> use for US letter paper
2: \usepackage[]{graphicx}
3: \usepackage{graphicx}% Include figure files
4: \usepackage{dcolumn}% Align table columns on decimal point
5: \usepackage{bm}% bold math
6:
7: \def\figsize{0.6}
8:
9: \begin{document}
10: \title{Transient current in a quantum dot asymmetrically
11: coupled to metallic leads}
12: \author{A. Goker$^1$, B. A. Friedman$^2$ and P. Nordlander$^1$}
13:
14: \address{$^1$
15: Department of Physics and Department of Electrical and Computer Engineering, \\
16: Rice Quantum Institute, Rice University, Houston, TX 77251-1892, USA
17: }
18: \address{$^2$
19: Department of Physics,
20: Sam Houston State University, Huntsville, TX 77341, USA
21: }
22: \date{\today}
23:
24: \ead{nordland@rice.edu}
25:
26: \begin{abstract}
27: The time-dependent non-crossing approximation is used to study
28: the transient current in a single electron transistor attached
29: asymmetrically to two leads following a sudden change in the
30: energy of the dot level. We show that for asymmetric coupling,
31: sharp features in the density of states of the leads can induce
32: oscillations in the current through the dot. These oscillations persist
33: to much longer timescales than the timescale for charge fluctuations.
34: The amplitude of the oscillations increases as the temperature or
35: source-drain bias across the dot is reduced and saturates for values
36: below the Kondo temperature. We discuss the microscopic origin of these
37: oscillations and comment on the possibility for their experimental detection.
38: \end{abstract}
39:
40: \pacs{72.15.Qm, 85.30.Vw, 73.50.Mx}
41:
42: \maketitle
43:
44: \section{Introduction}
45: Quantum effects are likely to play an increasing role in electronic devices as
46: the physical size of their components continues to shrink.
47: Quantum dots and qubits are examples of devices where quantum effects play a direct
48: role in their function. In a Single Electron Transistor (SET), i.e.,
49: a quantum dot coupled to two metallic leads, the conductance
50: can be drastically enhanced by the Kondo effect which can
51: occur when the quantum dot is populated by an odd number of
52: electrons \cite{NgLee88PRL,PustilnukGlazman04JPCM,GiulianoetAl04JPCM,%
53: GoldhaberetAl98Nature,CronenwettetAl98Science,HewsonetAl05JPCM,JiangSun07JPCM,%
54: GalperinetAl07JPCM}.
55: The Kondo effect is a quantum-coherent many-body state in which a spin
56: singlet state is formed between the unpaired localized electron and
57: delocalized electrons at the Fermi energy at low
58: temperatures \cite{Kondo64PTP}.
59:
60: An important issue for the function of any electronic device is how fast
61: it can respond to time-dependent perturbations and
62: bias \cite{ElzermanetAl04Nature,XingetAl07PRB}.
63: Several studies of time-dependent response of symmetrically-coupled SET
64: have been performed using a variety of methods
65: \cite{HettlerSchoeller95PRL,SchillerHershfield96PRL,%
66: GoldinAvishai98PRL,NordlanderetAl99PRL,KaminskyetAl99PRL,%
67: SchillerHershfield00PRB,KoganetAl04Science,LobaskinKehrein05PRB,%
68: MonrealFlores05PRB,GalperinetAl07PRB}.
69: In the case of a sudden switching of the dot level, the transient current
70: has been found to exhibit several distinct timescales
71: \cite{MerinoMarston04PRB,PlihaletAl05PRB,AndersSchiller05PRL,AndersetAl06PRB}.
72: The fastest timescale is associated with charge relaxation and the other
73: much slower timescales are associated with the formation of a Kondo state,
74: i.e. spin dynamics. The detailed evolution of
75: the instantaneous currents following a sudden change of the dot level
76: has been shown to depend sensitively on external parameters such
77: as source-drain bias, external temperature, dot-lead coupling
78: and position of the dot level. These above mentioned time-dependent
79: studies have all
80: been concerned with quantum dots which are symmetrically coupled to
81: their leads.
82: While the effect of asymmetric coupling on the steady-state conductance of an SET
83: has been studied both theoretically and
84: experimentally,\cite{KrawiecWysokinski02PRB,QuirionetAl06EPJB}
85: to our knowledge, the transient transport properties of an asymmetrically
86: coupled dot in the Kondo regime have not been investigated previously.
87:
88: In this paper, we use a recently developed multi-scale
89: many-body transport method to study the effect
90: of asymmetric dot-lead coupling on the
91: transient transport in a quantum dot \cite{IzmaylovetAl06JPCM}.
92: We show that for a quantum dot asymmetrically coupled to two leads
93: with sharp features in their Density of States (DOS),
94: the current can display sinusoidal modulations for timescales
95: well beyond the fast charge relaxation timescale.
96: The frequency of these sinusoidal modulations is given by the
97: energy difference between the Kondo resonance and the DOS feature.
98: The amplitude of the oscillations is found to increase
99: with decreasing temperature and source-drain bias and saturate for temperatures
100: below the Kondo scale.
101: We attribute this phenomenon to an interference effect between
102: the Kondo resonance at the Fermi level of the leads and the conduction
103: electrons around the DOS feature.
104: The magnitude of the oscillations depends sensitively on the
105: structure of the DOS feature of the leads. We show that
106: these oscillations also can occur for leads with a smooth DOS
107: but with finite bandwidth.
108:
109: \section{Time dependent current in infinite-U Anderson model}
110:
111: The SET is modeled as a single spin degenerate level of
112: energy $\varepsilon_{dot}$ coupled to leads through tunnel barriers,
113: \begin{eqnarray}
114: H(t)&=&\sum_{\sigma} \varepsilon_{dot}(t)n_{\sigma}
115: +\sum_{k\alpha\sigma}\varepsilon_{k}n_{k\alpha\sigma}
116: +{\textstyle\frac{1}{2}}{\sum} U n_{\sigma}n_{\sigma'}+ \nonumber \\
117: & &\sum_{k\alpha\sigma} \left[ V_{\alpha}(\varepsilon_{k\alpha})c_{k\alpha\sigma}^{\dag}c_{\sigma}+
118: {\rm h.c.} \right],
119: \label{Anderson}
120: \end{eqnarray}
121: where $c^\dagger_\sigma$ ($c_\sigma$) and $c^\dagger_{k\alpha\sigma}$
122: ($c_{k\alpha\sigma}$) with $\alpha$=L,R create (annihilate) an electron
123: of spin $\sigma$ in the dot level and in the left(L) and right(R) leads
124: respectively. The $n_\sigma$ and $n_{k\alpha\sigma}$ are the corresponding
125: number operators and $V_{\alpha}$ are the hopping amplitudes for the left
126: and right leads. The Coulomb repulsion energy $U$ is assumed to be sufficiently
127: large that double occupancy of the dot level is prohibited. In the following,
128: we will use atomic units with $\hbar=k_{\rm B}=e=1$.
129:
130: To deal with large Coulomb correlation, we
131: use the auxiliary boson method for the Anderson Hamiltonian.
132: Briefly, the ordinary electron operators on the site are decomposed
133: into a massless boson operator and a pseudofermion operator as
134: \begin{equation}
135: c_{\sigma}=b^{\dagger} f_{\sigma},
136: \end{equation}
137: with the constraint that the number of massless "slave" bosons
138: plus the number of pseudofermions
139: \begin{equation}
140: Q=b^{\dagger}b+\sum_{\sigma}f^{\dagger}_{\sigma}f_{\sigma}
141: \end{equation}
142: must be equal to unity. Self-energies for the pseudofermion and massless
143: "slave" boson are then projected onto the physically relevant $Q=1$
144: subspace.
145:
146: The Green's functions of the dot levels are calculated using the Non Crossing
147: Approximation (NCA) by numerical integration on a multi-scale time
148: grid \cite{IzmaylovetAl06JPCM}. When the dot
149: level $\varepsilon_{dot}$ lies well below the Fermi level $\varepsilon_{F}$,
150: the spectral function of the dot (local density of states) exhibit two features:
151: a broad Fano-like resonance of full-width
152: \begin{equation}
153: \Gamma_{tot}(\varepsilon) =
154: 2 \pi \sum_{k} (|V_{L} (\varepsilon_k)|^2 +|V_{R} (\varepsilon_k)|^2)
155: \delta(\varepsilon-\varepsilon_{k})
156: \label{gamma}
157: \end{equation}
158: around the dot level, and
159: a sharp temperature sensitive
160: resonance at the Fermi level (the Kondo peak), characterized by a low
161: energy scale $T_K$ (the Kondo temperature)
162: \begin{equation}
163: T_K \propto \left(\frac{D\Gamma_{tot}}{4}\right)^\frac{1}{2}
164: \exp\left(-\frac{\pi|\varepsilon_{\rm dot}|}{\Gamma_{tot}}\right),
165: \label{tkondo}
166: \end{equation}
167: where $D$ is a high energy cutoff equal to half bandwidth
168: of the conduction electrons and $\Gamma_{tot}$ corresponds to
169: the value of $\Gamma_{tot}(\varepsilon)$ at $\varepsilon=\varepsilon_F$.
170: All energy units in this paper will be given in terms of $\Gamma_{tot}$.
171:
172: The current flowing through the SET can be calculated as the difference between
173: the currents from the left and right leads as
174: \begin{equation}
175: I(t)=I_L(t)-I_R(t),
176: \label{eq:ILR}
177: \end{equation}
178: The most general expression for the net current flowing from
179: the left(right) lead is given by \cite{JauhoetAl94PRB}
180: \begin{eqnarray}
181: I_{L(R)}(t)&=&-2Im( \int_{-\infty}^{t} dt_1 \int \frac{d\varepsilon}{2\pi}
182: e^{-i\varepsilon(t_1-t)}\Gamma_{L(R)}(\varepsilon) \nonumber \\
183: & & e^{i\int_{t_1}^{t}d\tau \Delta_{L(R)}(\tau)}
184: [G^{<}(t,t_1)+f_{L(R)}(\varepsilon)G^{R}(t,t_1)])
185: \label{wingreen}
186: \end{eqnarray}
187: where the coupling functions $\Gamma_{L(R)}$
188: \begin{equation}
189: \Gamma_{L(R)}(\varepsilon)=2 \pi \rho_{L(R)}(\varepsilon)
190: V_{L(R)}(\varepsilon) V_{L(R)}^{*}(\varepsilon),
191: \end{equation}
192: depend on the DOS of the leads $\rho_{L(R)}(\varepsilon)$ and the hopping matrix
193: elements in Eq.(\ref{Anderson}). The quantity
194: $\Delta_{L(R)}$ represents the time-dependence of the
195: single particle energies in the left and right leads.
196: In this paper, we consider the case of a small constant
197: bias across the leads and no explicit time and energy dependence of the hopping
198: matrix elements, therefore $V_{L(R)}(\varepsilon)=V_{L(R)}(\varepsilon_f)$.
199: We further make the assumption that the DOS of the leads
200: are the same, i.e. $\rho_{L(R)}(\varepsilon)=\rho(\varepsilon)$.
201: Thus the coupling functions can be parameterized
202: as,
203: \begin{equation}
204: \Gamma_{L(R)}(\varepsilon)=\bar{\Gamma}_{L(R)} \rho(\varepsilon)
205: \end{equation}
206: where $\bar{\Gamma}_{L(R)}$ are constants given by
207: $\bar{\Gamma}_{L(R)}=2 \pi |V_{L(R)}(\varepsilon_f)|^2$ and
208: they determine the asymmetry of the couplings. In terms of
209: these constants,
210: $\Gamma_{tot}=(\bar{\Gamma}_{L}+\bar{\Gamma}_{R})\rho(\varepsilon_f)$.
211:
212: In Fig.~1 we show the functions $\rho(\varepsilon)$
213: that will be used to model the leads.
214: The Lorentzians used in Fig.~1c and Fig.~1d to model sharp DOS features
215: have a width equal to 0.002~$\Gamma_{tot}$.
216: The DOS function $\rho(\varepsilon)$ has been normalized for all cases such
217: that the bands contain the same number of electrons.
218: The bandwidth of the leads is assumed to be 2$D$
219: with the Fermi energy at $\varepsilon=0$.
220:
221: \begin{figure}[htb]
222: \centerline{\includegraphics[width=12cm]{fig1.eps}}
223: \caption{
224: This figure shows the functions $\rho(\varepsilon)$
225: used to model the leads.
226: Panel a) corresponds to a rectangular band, b) to a parabolic band,
227: and c) and d) to a parabolic band with a Lorentzian feature
228: located at $\varepsilon$=-D/2
229: and $\varepsilon$=-3D/4 respectively
230: }
231: \end{figure}
232:
233: The expression for the current can then be written as
234: \begin{eqnarray}
235: I(t)&=&-2\bar{\Gamma}_{L}
236: Im(\int_{-\infty}^{t} dt_1 (G^{<}(t,t_1)h(t-t_1)+G^{R}(t,t_1)f_{L}(t-t_1))) \nonumber \\
237: & &+2\bar{\Gamma}_{R}Im(\int_{-\infty}^{t} dt_1
238: (G^{<}(t,t_1)h(t-t_1)+G^{R}(t,t_1)f_{R}(t-t_1))),
239: \label{current1}
240: \end{eqnarray}
241: where
242: \begin{equation}
243: h(t-t_1)=\int_{-D}^{D} \frac{d\varepsilon}{2\pi}\rho(\varepsilon)e^{i\varepsilon(t-t_1)},
244: \label{h}
245: \end{equation}
246: \begin{equation}
247: f_{L}(t-t_1)=\int_{-D}^{D} \frac{d\varepsilon}{2\pi}\rho(\varepsilon)
248: \frac{e^{i\varepsilon(t-t_1)}}{1+e^{\beta \left [\varepsilon-\frac{V}{2}\right]}},
249: \label{f}
250: \end{equation}
251: and
252: \begin{equation}
253: f_{R}(t-t_1)=\int_{-D}^{D} \frac{d\varepsilon}{2\pi}\rho(\varepsilon)
254: \frac{e^{i\varepsilon(t-t_1)}}{1+e^{\beta \left [\varepsilon+\frac{V}{2}\right]}}.
255: \label{fr}
256: \end{equation}
257: In these expressions, $V$ represents the source-drain bias.
258:
259: The physical Green's functions in
260: Eq.~(\ref{current1}) in can be expressed in terms of the pseudofermion and slave
261: boson Green's functions using a projection approach discussed
262: previously \cite{ShaoetAl194PRB}.
263: The final expression for the current is
264: \begin{eqnarray}
265: I(t)&=&-2\bar{\Gamma}_{L}\textit{Im} (\int_{-\infty}^{t} dt_1
266: (G_{pseudo}^{<}(t,t_1)B^{R}(t_1,t)h(t-t_1)\nonumber \\
267: & &-i((G_{pseudo}^{R}(t,t_1)B^{<}(t_1,t)+G_{pseudo}^{<}(t,t_1)
268: B^{R}(t_1,t))f_{L}(t-t_1)))\nonumber \\
269: & &+2\bar{\Gamma}_{R}\textit{Im} (\int_{-\infty}^{t} dt_1
270: (G_{pseudo}^{<}(t,t_1)B^{R}(t_1,t)h(t-t_1)\nonumber \\
271: & &-i((G_{pseudo}^{R}(t,t_1)B^{<}(t_1,t)+G_{pseudo}^{<}(t,t_1)
272: B^{R}(t_1,t))f_{R}(t-t_1))).
273: \label{final}
274: \end{eqnarray}
275: This is the main result of this section and will be used
276: in the calculations presented below.
277:
278: \section{Results}
279:
280: In this section we will analyze the instantaneous current
281: following a sudden change of the dot level from a position
282: at $\varepsilon_1$=5$\Gamma_{tot}$ below the Fermi
283: level where, for the present finite temperatures, the Kondo effect will
284: be absent to a position at $\varepsilon_2$=2$\Gamma_{tot}$
285: closer to the Fermi energy where the Kondo effect will be present.
286: For a parabolic DOS function (Fig.~1b) with $D$=9$\Gamma_{tot}$,
287: the Kondo temperature in the final state is approximately
288: $T_K=$0.0016$\Gamma_{tot}$.
289:
290: \begin{figure}[htb]
291: \centerline{\includegraphics[width=12cm]{fig2.eps}}
292: \caption{
293: Black(dotted), blue(dashed) and red(dot dashed) curves in panel a show the
294: instantaneous conductance vs. time for rectangular DOS
295: in Fig. 1a immediately after the dot level is switched to its final position
296: for asymmetry factors of 0.95,0.9 and 0.85 respectively.
297: $\Gamma_{tot}$ is fixed with D=9$\Gamma_{tot}$ at T=0.0093$\Gamma_{tot}$
298: and V=$T_K$. The beginning of the oscillations is clearly visible in this panel.
299: Black(dotted), blue(dashed) and red(dot dashed) curves in panel b are the continuation
300: of the ones in the first panel in the long timescale for the same parameters.
301: }
302: \label{asymmetry}
303: \end{figure}
304:
305: We begin our analysis by investigating the effect of
306: asymmetric coupling to the leads. We define the asymmetry factor as the
307: the ratio $\bar{\Gamma}_{L}/\bar{\Gamma}_{tot}$
308: where $\bar{\Gamma}_{tot}$= $\bar{\Gamma}_{L}+\bar{\Gamma}_{R}$.
309: In Fig.~2, the instantaneous conductance, $G=I(t)/V$, is plotted as a function of time
310: after the dot level is switched for various asymmetry factors with small bias.
311: The final steady state conductances are in perfect agreement with
312: previous theoretical results \cite{WingreenMeir94PRB}.
313: The transient short timescale associated with $\Gamma_{tot}$ shown
314: in Fig.~2a is due to the formation of the broad Fano-like
315: resonance at $\varepsilon_2$. The transient increase in the instantaneous
316: current is due to the charging of the dot. The steady state current is determined by the asymmetry
317: factor. For an asymmetry factor of 1 where the steady state current is zero, we have verified the
318: accuracy of our numerical approach by showing that the integrated current is equal to the change of
319: the charge of the dot level.
320:
321: Fig.~2b shows the instantaneous conductances
322: for larger times on a magnified scale. It is clear from this panel that the
323: current exhibits sinusoidal modulations at timescales well beyond
324: $\Gamma_L$ or $\Gamma_R$. As we reduce the
325: asymmetry factor, the amplitude of these sinusoidal modulations
326: starts decreasing and eventually disappears for symmetric coupling.
327: The frequency of the conductance oscillations is
328: equal to the bandwidth $D$ of the leads. External parameters such as
329: the energy or width of the dot level, asymmetry factor, and ambient temperature
330: and source-drain bias only influence the amplitude of the oscillations.
331:
332: \begin{figure}[htb]
333: \centerline{\includegraphics[width=12cm]{fig3.eps}}
334: \caption{
335: Effect of the DOS function on the instantaneous conductance.
336: Red(dotted), dark blue(dot dashed), black(solid) and green(dashed) curves display
337: the instantaneous conductance vs. time in the long timescale for cases a, b, c and d
338: respectively in Fig.1 when the source-drain bias is equal to V=$T_K$
339: with fixed $\Gamma_{tot}$. All curves are for asymmetry factor of 0.9,
340: D=9$\Gamma_{tot}$ at T=0.0093$\Gamma_{tot}$.
341: }
342: \label{dos}
343: \end{figure}
344:
345: In Fig.~3 we show the effect of the the DOS function of the leads
346: $\rho(\epsilon)$ on the conductance oscillations.
347: The oscillations do not look like perfectly sinusoidal functions
348: because of our use of a finite time step in the numerical solution
349: of the Dyson equations. However, the results are fully converged
350: and for a finer time steps we recover almost perfectly sinusoidal oscillations.
351: Figure~3 reveals that the
352: DOS can have a pronounced effect on the conductance oscillations.
353: The largest oscillations occur for a DOS with a peak feature as
354: in Fig.~1c and 1d. For a parabolic DOS function (Fig.~1b),
355: conductance oscillations are still present but not discernible on the scale
356: of the figure. The frequency of the current oscillations is equal to
357: the energy difference between the Fermi level of the leads and the feature
358: in the DOS function. For the rectangular and parabolic DOS
359: function (Fig.~1a and 1b) where the DOS feature is the band cut-off,
360: the period is equal to 1/$D$=1/9 and for the DOS functions
361: depicted in Fig.~1c and 1d where the DOS feature is the narrow
362: Lorentzian peak, the periods are 2/$D$ and 4/3$D$ respectively.
363:
364: We have also investigated other shapes of DOS functions.
365: For instance, for a triangular DOS, $\rho(\varepsilon)=2(1+\varepsilon/D)$, the conductance
366: displays oscillations whose frequency is equal to $D$ very similar to
367: what was obtained for the parabolic DOS in Fig.~1b.
368: For two sharp features at different energies $D_1$ and $D_2$ in the DOS
369: we obtain conductance modulations of frequencies proportional
370: to $(D_1+D_2)$ and $(D_1-D_2)$.
371:
372: Figure.~4 shows the effect of temperature and source-drain
373: bias on the conductance oscillations. The figure demonstrates that
374: the amplitude of the oscillations decrease with temperature and
375: bias across the dot but their frequency remain unchanged.
376: The figure shows two important effects. First, it takes longer for the current
377: and thus the damped oscillations to decay as the temperature or
378: source-drain bias is reduced. Second, the amplitude
379: of the damped oscillations increases with decreasing temperature
380: or source-drain bias and but saturates for temperatures and bias below
381: the Kondo temperature, which we estimate to be
382: around $T_K=$0.0016$\Gamma_{tot}$ in these systems.
383:
384: \begin{figure}[htb]
385: \centerline{\includegraphics[width=12cm]{fig4.eps}}
386: \caption{
387: Panel a shows the instantaneous conductance for
388: several different temperatures with asymmetry factor of 0.9,
389: fixed $\Gamma_{tot}$, D=9$\Gamma_{tot}$ and rectangular
390: DOS. Dark blue(dot dot dashed), red(dot dashed), green(dashed),
391: light blue(dotted) and black(solid) curves represent the conductance at T=0.0186,
392: 0.0093, 0.0046, 0.0023, and 0.0009$\Gamma_{tot}$ respectively.
393: Red(dot dashed), black(dotted) and dark blue(dashed) curves in panel b display
394: the instantaneous conductance vs. time in the long timescale for rectangular
395: DOS when the source-drain bias is equal to V=$T_K$, V=5$T_K$ and
396: V=10$T_K$ respectively with fixed $\Gamma_{tot}$ and D=9$\Gamma_{tot}$
397: for asymmetry factor of 0.9 at T=0.0093$\Gamma_{tot}$.
398: }
399: \label{bias}
400: \end{figure}
401:
402: \section{Discussion}
403:
404:
405: Our numerical calculations clearly show that the timescale for the decay of the
406: conductance oscillations is much larger than the fast timescales set by the
407: couplings of the dot level to the leads, $\Gamma_L$ or $\Gamma_L$.
408: The timescale does not depend on the width or shape of the DOS feature in the
409: leads. It appears that the timescale is related to the Kondo resonance.
410: When the Kondo resonance is fully formed at times around $1/T_K$, the oscillations
411: disappears.
412:
413: Further support for the role of Kondo physics in these conductance oscillations
414: comes from the observation of a
415: saturation of their amplitude for temperatures
416: and source-drain bias below the Kondo temperature (Fig.~4).
417: The effect of temperature and source-drain bias on the Kondo resonance in an SET
418: was recently investigated in detail \cite{PlihaletAl05PRB}. For values smaller than
419: the Kondo temperature, the Kondo resonance is fully formed just above the Fermi Level.
420: Increasing the temperature above $T_K$ broadens and reduces the magnitude of the
421: resonance. Increasing the source-drain bias results in the formation of a split Kondo
422: resonance with strongly reduced intensities. For temperatures or source-drain
423: bias well above the Kondo temperature, the Kondo resonance is completely
424: suppressed. Since the frequency of the
425: oscillations is determined by the energy difference between the Fermi level and
426: the DOS feature, we believe that the oscillations reflect an
427: interference process between the conduction electrons associated with the Kondo
428: resonance and the conduction electrons associated with the DOS feature.
429: In order to substantiate this explanation we used a simple analytical
430: noninteracting Anderson model to calculate the instantaneous
431: conductance following the sudden switching of a dot level of a width $\Gamma_{eff}$
432: equal to that of the final Kondo resonance in the interacting model.
433: This effective resonance was switched to a position just above the Fermi level
434: of the leads. These calculations were performed numerically
435: for both the parabolic and rectangular DOS functions shown in Fig.~1a and 1b.
436: The instantaneous
437: conductance was found to display the same conductance oscillations with frequencies
438: determined by the bandwidth $D$ as in the original interacting model. The oscillations
439: are found to decay over a long timescale determined by $\Gamma_{eff}$.
440:
441: While the Kondo temperature depends only on the total coupling to the
442: leads $\Gamma_{tot}$, the total current depends on the asymmetry factor.
443: The reason why the conductance oscillations only show up for asymmetric
444: couplings is that the total current through the dot
445: is the difference of the currents from the left and right leads Eq.~(\ref{eq:ILR}).
446: The interference effect shows up both in the the right and left currents.
447: For symmetric coupling, the current oscillations from the right and left leads
448: are out of phase resulting in a cancellation of the oscillations
449: in the total current. For asymmetric coupling the left and right current oscillations
450: do not cancel resulting in the observed conductance oscillations.
451: This result has also been verified numerically for the noninteracting model
452: described in the preceding paragraph.
453:
454: For dots coupled to leads with very weak DOS features,
455: the conductance oscillations will be very small. In Fig.~5 we show the
456: time-derivative of the instantaneous current for leads with a
457: parabolic DOS function (Fig.~1b) for three different values
458: of the bandwidth $D$. The conductance oscillations here
459: results from the interference of the Kondo resonance in the final state
460: and the weak discontinuity in the DOS at the lower band edge
461: of the leads. The figure clearly demonstrates oscillations of the instantaneous
462: currents with a frequency equal to $D$ in the long timescale.
463:
464: \begin{figure}[htb]
465: \centerline{\includegraphics[width=12cm]{fig5.eps}}
466: \caption{
467: Time-derivative of the instantaneous current in the long timescale
468: for parabolic DOS (Fig.~1b). Black(solid), red(dashed)
469: and dark blue(dot dashed) curves correspond to D=6.75$\Gamma_{tot}$,
470: D=9$\Gamma_{tot}$ and D=13.5$\Gamma_{tot}$ for asymmetry
471: factor of 1.0 at T=0.0093$\Gamma_{tot}$.
472: }
473: \end{figure}
474:
475: The experimental study of the conductance oscillations could possibly
476: be made using previously suggested techniques, i.e., by measuring the total
477: charge transport as function of pulse duration \cite{NordlanderetAl99PRL}.
478: For a bandwidth of $D$=1~eV, the oscillation period will be of
479: an order of $10^{-14}$ seconds. Since the current oscillates, electromagnetic
480: radiation will be emitted at a rate proportional to $[\frac{dI(t)}{dt}]^2$.
481: For a suitably designed system it may be possible to detect the emitted light.
482: For $D$=1~eV the emission would occur in the
483: infrared at a photon energy of 0.1~eV.
484:
485:
486: \section{Conclusion}
487: In this paper, we employed the time dependent non-crossing approximation
488: to analyze the transient current in a single electron transistor in the
489: Kondo regime asymmetrically coupled to two metallic leads with features
490: in their DOS.
491: We show that for asymmetric coupling, the conductance can exhibit oscillations
492: which persist for times much longer than the timescale for charge relaxation.
493: The origin of these oscillations is an interference between the conduction
494: electrons associated with the Kondo resonance and those associated with the
495: DOS feature. The amplitude of the oscillations are found to depend strongly on
496: temperature and source-drain bias when those exceed the Kondo temperature.
497: We hope that our
498: predictions will motivate further theoretical and experimental studies.
499:
500:
501: \section*{Acknowledgments}
502: We thank Prof. Hong Guo for communication
503: regarding relevant work \cite{MaciejkoetAl06PRB}.
504: This work was supported by the Welch Foundation under grant C-1222.
505:
506: \clearpage
507:
508:
509: \providecommand{\newblock}{}
510: \begin{thebibliography}{10}
511: \expandafter\ifx\csname url\endcsname\relax
512: \def\url#1{{\tt #1}}\fi
513: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
514: \providecommand{\eprint}[2][]{\url{#2}}
515: % Bibliography created with iopart-num.bst, v1.0
516:
517: \bibitem{NgLee88PRL}
518: Ng T~K and Lee P~A 1988 {\em Phys. Rev. Lett.\/} {\bf 61} 1768
519:
520: \bibitem{PustilnukGlazman04JPCM}
521: Pustilnik M and Glazman L 2004 {\em J. Phys.:Condens. Matter\/} {\bf 16}
522: R513--R537
523:
524: \bibitem{GiulianoetAl04JPCM}
525: Giuliano D, Naddeo A and Tagliacozzo A 2004 {\em J. Phys.:Condens. Matter\/}
526: {\bf 16} S1453--S1483
527:
528: \bibitem{GoldhaberetAl98Nature}
529: Goldhaber-Gordon D, Shtrikman H, Mahalu D, Abusch-Magder D, Meirav U and
530: Kastner M~A 1998 {\em Nature\/} {\bf 391} 156--159
531:
532: \bibitem{CronenwettetAl98Science}
533: Cronenwett S~M, Osterkamp T~H and Kouwenhoven L~P 1998 {\em Science\/} {\bf
534: 281} 540--544
535:
536: \bibitem{HewsonetAl05JPCM}
537: Hewson A~C, Bauer J and Oguri A 2005 {\em J. Phys.:Condens. Matter\/} {\bf 17}
538: 5413--5422
539:
540: \bibitem{JiangSun07JPCM}
541: Jiang Z~T and Sun Q~F 2007 {\em J. Phys.: Condens. Matter\/} {\bf 19} 156213
542:
543: \bibitem{GalperinetAl07JPCM}
544: Galperin M, Ratner M~A and Nitzan A 2007 {\em J. Phys.: Condens. Matter\/} {\bf
545: 19} 103201
546:
547: \bibitem{Kondo64PTP}
548: Kondo J 1964 {\em Prog. Theor. Phys.\/} {\bf 32} 37
549:
550: \bibitem{ElzermanetAl04Nature}
551: Elzerman J~M, Hanson R, van Beveren L~H~W, Witkamp B, Vandersypen L~M~K and
552: Kouwenhoven L~P 2004 {\em Nature(London)\/} {\bf 430} 431--434
553:
554: \bibitem{XingetAl07PRB}
555: Xing Y, Sun Q~F and Wang J 2007 {\em Phys. Rev. B\/} {\bf 75} 125308
556:
557: \bibitem{HettlerSchoeller95PRL}
558: Hettler M~H and Schoeller H 1995 {\em Phys. Rev. Lett.\/} {\bf 74} 4907
559:
560: \bibitem{SchillerHershfield96PRL}
561: Schiller A and Hershfield S 1996 {\em Phys. Rev. Lett.\/} {\bf 77} 1821
562:
563: \bibitem{GoldinAvishai98PRL}
564: Goldin Y and Avishai Y 1998 {\em Phys. Rev. Lett.\/} {\bf 81} 5394
565:
566: \bibitem{NordlanderetAl99PRL}
567: Nordlander P, Pustilnik M, Meir Y, Wingreen N~S and Langreth D~C 1999 {\em
568: Phys. Rev. Lett.\/} {\bf 83} 808--811
569:
570: \bibitem{KaminskyetAl99PRL}
571: Kaminsky A, Nazarov Y~V and Glazman L~I 1999 {\em Phys. Rev. Lett.\/} {\bf 83}
572: 384
573:
574: \bibitem{SchillerHershfield00PRB}
575: Schiller A and Herschfield S 2000 {\em Phys. Rev. B\/} {\bf 62} R16271--R16274
576:
577: \bibitem{KoganetAl04Science}
578: Kogan A, Amasha S and Kastner M~A 2004 {\em Science\/} {\bf 304} 1293
579:
580: \bibitem{LobaskinKehrein05PRB}
581: Lobaskin D and Kehrein S 2005 {\em Phys. Rev. B\/} {\bf 71} 193303
582:
583: \bibitem{MonrealFlores05PRB}
584: Monreal R~C and Flores F 2005 {\em Phys. Rev. B\/} {\bf 72} 195105
585:
586: \bibitem{GalperinetAl07PRB}
587: Galperin M, Nitzan A and Ratner M~A 2007 {\em Phys. Rev. B\/} p 035301
588:
589: \bibitem{MerinoMarston04PRB}
590: Merino J and Marston J~B 2004 {\em Phys. Rev. B\/} {\bf 69} 115304
591:
592: \bibitem{PlihaletAl05PRB}
593: Plihal M, Langreth D~C and Nordlander P 2005 {\em Phys. Rev. B\/} {\bf 71}
594: 165321
595:
596: \bibitem{AndersSchiller05PRL}
597: Anders F~B and Schiller A 2005 {\em Phys. Rev. Lett.\/} {\bf 95} 196801
598:
599: \bibitem{AndersetAl06PRB}
600: Anders F~B and Schiller A 2006 {\em Phys. Rev. B\/} {\bf 74} 245113
601:
602: \bibitem{KrawiecWysokinski02PRB}
603: Krawiec M and Wysokinski K~I 2002 {\em Phys. Rev. B\/} {\bf 66} 165408
604:
605: \bibitem{QuirionetAl06EPJB}
606: Quirion Q, Weiss J and v~Klitzing K 2006 {\em Eur. Phys. J. B\/} {\bf 51}
607: 413--419
608:
609: \bibitem{IzmaylovetAl06JPCM}
610: Izmaylov A~F, Goker A, Friedman B~A and Nordlander P 2006 {\em J. Phys.:
611: Condens. Matter\/} {\bf 18} 8995--9006
612:
613: \bibitem{JauhoetAl94PRB}
614: Jauho A~P, Wingreen N~S and Meir Y 1994 {\em Phys. Rev. B\/} {\bf 50} 5528
615:
616: \bibitem{ShaoetAl194PRB}
617: Shao H~X, Langreth D~C and Nordlander P 1994 {\em Phys. Rev. B\/} {\bf 49}
618: 13929--13947
619:
620: \bibitem{WingreenMeir94PRB}
621: Wingreen N~S and Meir Y 1994 {\em Phys. Rev. B\/} {\bf 49} 11\,040
622:
623: \bibitem{MaciejkoetAl06PRB}
624: Maciejko J, Wang J and Guo H 2006 {\em Phys. Rev. B\/} {\bf 74} 085324
625:
626: \end{thebibliography}
627:
628:
629: \end{document}
630:
631: