1: %\documentstyle[preprint,aps]{revtex}
2: %\documentclass[aps,prbbib,twocolumn,epsf]{revtex4}
3: %\usepackage{picture}
4: %\usepackage{graphicx} % Include figure files
5: % Align table columns on decimal point
6: % bold math
7: %\input{tcilatex}
8:
9: \documentclass[aps,prb,showpacs,preprint,amsfonts,amssymb,amsmath]{revtex4}
10: %\documentclass[aps,prb,showpacs,groupedaddress,twocolumn]{revtex4}
11: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
12: \usepackage{amsmath,amssymb}
13: \usepackage{dcolumn}
14: \usepackage{bm}
15:
16: \setcounter{MaxMatrixCols}{10}
17: %TCIDATA{OutputFilter=LATEX.DLL}
18: %TCIDATA{Version=4.00.0.2321}
19: %TCIDATA{LastRevised=Tuesday, January 29, 2008 15:59:36}
20: %TCIDATA{<META NAME="GraphicsSave" CONTENT="32">}
21:
22: %\input{tcilatex}
23:
24: \begin{document}
25:
26: \title{Double quantum dot as detector of spin bias}
27: \author{Qing-feng Sun$^{1}$, Yanxia Xing$^{1}$, and Shun-Qing Shen$^{2}$}
28: \affiliation{$^1$Beijing National Lab for Condensed Matter Physics
29: and Institute of Physics, Chinese Academy of Sciences, Beijing
30: 100080, China
31: \\
32: $^2$Department of Physics, and Center of Theoretical and
33: Computational Physics, The University of Hong Kong, Hong Kong, China
34: }
35: \date{\today}
36:
37: \begin{abstract}
38: It was proposed that a double quantum dot can be used to be a
39: detector of spin bias. Electron transport through a double quantum
40: dot is investigated theoretically when a pure spin bias is applied
41: on two conducting leads contacted to the quantum dot. It is found
42: that the spin polarization in the left and right dots may be
43: induced spontaneously while the intra-dot levels are located
44: within the spin bias window and breaks the left-right symmetry of
45: the two quantum dots. As a result, a large current emerges. For an
46: open external circuit an charge bias instead of a charge current
47: will be induced in equilibrium, which is believed to be measurable
48: according to the current nanotechnology. This method may provide a
49: practical and whole electrical approach to detect the spin bias
50: (or the spin current) by measuring the charge bias or current in a
51: double quantum dot.
52: \end{abstract}
53:
54: \pacs{85.75.-d, 85.35.-p, 73.21.La, 73.23.-b}
55: \maketitle
56:
57: \section{Introduction}
58:
59: Discovery and application of giant magnetoresistance (GMR) in metallic thin
60: films marks the beginning of a new era of spintronics.\cite{ref1,ref2} Since
61: then, people begin to exploit electron spin to replace the role of electron
62: charge in electronic devices. As a counterpart of charge current, spin
63: current, in which spin-up and spin-down electrons move coherently in
64: opposite directions, has been attracted extensive interests.\cite{Awschalom}
65: Various methods were proposed to generate spin current,\cite{ref3} and to
66: explore the characteristics of the spin transport. Over last few years,
67: search of spin current has made a great of progresses. It has been generated
68: and detected successfully by various means, such as the optical injection,%
69: \cite{ref4, cui} the magnetic tunnelling injection,\cite{ref5,Kimura} or the
70: spin Hall effect.\cite{ref6,ref7} All these experiments focus on the optical
71: measurement of spin accumulation near the boundaries of sample or electric
72: measurement of the scattering effect induced by the spin current via
73: spin-orbital coupling. There are also some proposals to measure spin current
74: or spin polarized current,\cite{ref8,addref1,ref9,add1} e.g. to measure the
75: spin torque while a spin current flowing through a ferromagnetic-nonmagnetic
76: interface,\cite{ref8} or to detect the induced electric field by the spin
77: current.\cite{addref1,ref9} In all these methods, it always involves the
78: optical, magnetic materials or impurities, magnetic field, or spin-orbit
79: interaction. Up to now, it is still a challenge to detect the spin current
80: efficiently, which has become a bottleneck of the development of the
81: spintronics.
82:
83: When a spin current flows through a device, there always exists a spin bias
84: between the two terminals of the device.\cite{addnote1} A spin bias means
85: that the chemical potentials of the two terminals are spin-dependent (see
86: Fig.1). The spin bias is regarded as the driving force behind the spin
87: current. When the circuit is open, the spin current has to be zero.
88: Consequently the spin bias usually induce spin accumulation in equilibrium.
89: When the circuit is connected, a spin current circulates. The relation
90: between spin bias and spin current is very similar with the relation between
91: the charge bias and charge current. On the charge transport, people often
92: detect the charge bias to replace the measurement of the charge current.
93: Correspondingly we can also measure the spin bias instead of the spin
94: current. In this paper, we propose an effective method to detect the spin
95: bias.
96:
97: The present proposal is a whole electric measurement of spin bias by means
98: of a double quantum dot (DQD). It does not involve any optical or magnetic
99: means, and even the spin-orbit interaction. The spin bias can be detected by
100: measuring the (charge) bias. The DQD can be regarded as an artificial
101: molecule, and the electron numbers in DQD can be controlled very well. In
102: last two decades, the electron transport through the DQD device has been
103: extensively investigated.\cite{ref10,ref11} DQD has also been proposed as a
104: qubit,\cite{ref12} a device to detect various tunnelling rates and spin flip
105: rate,\cite{ref11,ref13} and so on. Here we propose that a DQD can be applied
106: to measure the spin bias or spin current.
107:
108: Let us first describe the working mechanism of DQD as a detector of spin
109: bias. Consider a DQD coupled into two conducting leads. Suppose a spin bias
110: be applied between the left and right leads. Our task is to measure this
111: spin bias \textsl{experimentally}. The spin bias is defined as the
112: spin-dependent chemical potentials of the two leads with $\mu _{L\uparrow
113: }=-\mu _{L\downarrow }=-\mu _{R\uparrow }=\mu _{R\downarrow }=V$ (see Fig.1).%
114: \cite{ref14} Assume that the left-dot level $\epsilon _{L}$ is set at zero
115: and the right-dot level $\epsilon _{R}$ is at $-U$, where $U$ is the
116: intra-dot electron-electron (e-e) Coulomb interaction. This particular level
117: position is chosen to demonstrate the physics in our proposal, and is not
118: necessary at all in a general case. The left dot has a spin-up electron
119: because of $\mu _{L\uparrow }>\epsilon _{L}>\mu _{L\downarrow }$, while the
120: right dot, because of $\mu _{R\downarrow }>\epsilon _{R}+U>\mu _{R\uparrow
121: }>\epsilon _{R}$, is occupied by a spin-down electron, and its spin-up level
122: is consequently pushed away to the higher energy $\epsilon _{R}+U$ and is
123: empty (see Fig.1). The spin-up electron can then tunnel from the left lead
124: via the two dots to the right lead (see Fig.1a). Oppositely the spin-down
125: electron can hardly flow from the right lead to the left lead because of the
126: Pauli exclusion principle and the occupancy of the spin-down level in the
127: right dot (see Fig.1b). This breaks the symmetry of the motion of spin-up
128: and spin-down electrons in a pure spin bias. As a result, a (charge) current
129: circulates. This induced current can be measured experimentally, and
130: consequently be applied to measure the spin bias.
131:
132: The paper is organized as follows. In Section II, the model for the
133: DQD and the general formalism for nonequilibrium Keldysh Green's
134: function method are presented. The spin-bias-induced charge current
135: $J$ and the electron occupation numbers in the DQD are calculated.
136: In Section III, we take the numerical investigation. The
137: spin-dependent charge stability diagram in terms of the spin bias is
138: obtained. In Section IV, the induced charge bias in an open circuit
139: is numerically studied. Finally, a brief summary is presented in
140: Section V.
141:
142: \section{model and formulation}
143:
144: In this section, we present the model Hamiltonian of this DQD and the
145: general formalism of Keldysh Green's function technique for electron
146: transport through the DQD. The DQD device is modelled by the following
147: Hamiltonian,
148: \begin{eqnarray}
149: H &=&\sum\limits_{\alpha ,k,\sigma }\epsilon _{\alpha k}a_{\alpha k\sigma
150: }^{\dagger }a_{\alpha k\sigma }+\sum\limits_{\alpha ,\sigma }\epsilon
151: _{\alpha }d_{\alpha \sigma }^{\dagger }d_{\alpha \sigma } \notag \\
152: &&+\sum\limits_{\alpha }U_{in}d_{\alpha \uparrow }^{\dagger }d_{\alpha
153: \uparrow }d_{\alpha \downarrow }^{\dagger }d_{\alpha \downarrow
154: }+\sum\limits_{\sigma ,\sigma ^{\prime }}U_{ex}d_{L\sigma }^{\dagger
155: }d_{L\sigma }d_{R\sigma ^{\prime }}^{\dagger }d_{R\sigma ^{\prime }} \notag
156: \\
157: &&+\sum\limits_{\alpha ,k,\sigma }t_{\alpha }a_{\alpha k\sigma }^{\dagger
158: }d_{\alpha \sigma }+\sum_{\sigma }t_{c}d_{L\sigma }^{\dagger }d_{R\sigma
159: }+H.c.
160: \end{eqnarray}%
161: where $a_{\alpha k\sigma }^{\dagger }$ ($a_{\alpha k\sigma }$) and $%
162: d_{\alpha \sigma }^{\dagger }$ ($d_{\alpha \sigma }$) are the creation
163: (annihilation) operators of electron with spin $\sigma $($=\uparrow
164: ,\downarrow )$ in the lead $\alpha $($=L,R)$ and the dot $\alpha $ ,
165: respectively. Each dot has a single energy level $\epsilon _{\alpha }$ and
166: an intra-dot e-e interaction $U_{in}$. In addition, the inter-dot e-e
167: interaction $U_{ex}$ is also included. We emphasize that the system does not
168: break the spin SU(2) symmetry, and the hopping coefficients $t_{\alpha }$
169: and $t_{c}$ are spin-independent.
170:
171: Following the transport theory of Keldysh Green's function,\cite{ref15} the
172: electron current $J_{\alpha \sigma }$ with the spin $\sigma $ from the lead $%
173: \alpha $ flowing into the dot $\alpha $ and the occupation number of
174: electron $n_{\alpha \sigma }$ at the level $\alpha ,\sigma $ can be
175: expressed as,
176: \begin{eqnarray}
177: J_{\alpha \sigma } &=&-Im\int \frac{d\epsilon }{2\pi }\Gamma _{\alpha }\left[
178: 2f_{\alpha \sigma }G_{\alpha \alpha \sigma }^{r}(\epsilon )+G_{\alpha \alpha
179: \sigma }^{<}(\epsilon )\right] \\
180: n_{\alpha \sigma } &=&\langle d_{\alpha \sigma }^{\dagger }d_{\alpha \sigma
181: }\rangle =-i\int \frac{d\epsilon }{2\pi }G_{\alpha \alpha \sigma
182: }^{<}(\epsilon )
183: \end{eqnarray}%
184: where $\Gamma _{\alpha }=2\pi \sum_{k}|t_{\alpha }|^{2}\delta (\epsilon
185: -\epsilon _{\alpha k})$. $f_{\alpha \sigma }(\epsilon )=1/\{exp[(\epsilon
186: -\mu _{\alpha \sigma })/k_{B}T]+1\}$ is the Fermi-Dirac distribution of
187: electrons in the leads. Because of the spin bias in the two leads, the
188: chemical potentials for spin-up and spin-down electrons are not equal. $%
189: G_{\alpha \alpha \sigma }^{r}(\epsilon )$ and $G_{\alpha \alpha \sigma
190: }^{<}(\epsilon )$ in Eqs. (2) and (3) are the standard retarded and the
191: Keldysh Green's functions of the QDs, they are the Fourier transformation of
192: $G_{\alpha \alpha \sigma }^{r,<}(t)$, where
193: \begin{eqnarray*}
194: G_{\alpha \alpha ^{\prime }\sigma }^{r}(t) &\equiv &-i\theta (t)\langle
195: \{d_{\alpha \sigma }(t),d_{\alpha ^{\prime }\sigma }^{\dagger }(0)\}\rangle ,
196: \\
197: G_{\alpha \alpha ^{\prime }\sigma }^{<}(t) &\equiv &i\langle d_{\alpha
198: \sigma }^{\dagger }(0)d_{\alpha ^{\prime }\sigma }(t)\rangle .
199: \end{eqnarray*}
200:
201: We first solve the Green's functions $\mathbf{g}_{\sigma }^{r}(\epsilon )$
202: of the isolated DQDs system (i.e. $t_{\alpha }=t_{c}=0$). Consider that the
203: spin bias $V$ is less than the intra-dot e-e interaction $U_{in}$ and the
204: two-electron co-tunneling events can be ignored. $\mathbf{g}_{\sigma
205: }^{r}(\epsilon )$ are obtained from the equation of motion technique:\cite%
206: {note21}
207: \begin{eqnarray}
208: g_{\alpha \alpha \sigma }^{r}(\epsilon ) &=&\frac{(1-n_{\alpha \bar{\sigma}%
209: })(1-\{n_{\bar{\alpha}}\})}{A}+\frac{(1-n_{\alpha \bar{\sigma}})\{n_{\bar{%
210: \alpha}}\}}{A-U_{ex}} \notag \\
211: &&+\frac{n_{\alpha \bar{\sigma}}(1-\{n_{\bar{\alpha}}\})}{A-U_{in}}+\frac{%
212: n_{\alpha \bar{\sigma}}\{n_{\bar{\alpha}}\}}{A-U_{in}-U_{ex}},
213: \end{eqnarray}%
214: and $g_{LR\sigma }^{r}=g_{RL\sigma }^{r}=0$, where $\bar{\alpha}=R$ for $%
215: \alpha =L$ and $\bar{\alpha}=L$ for $\alpha =R$, $\bar{\sigma}=\downarrow $
216: for $\sigma =\uparrow $ and $\bar{\sigma}=\uparrow $ for $\sigma =\downarrow
217: $, $A\equiv \epsilon -\epsilon _{\alpha }-[n_{\bar{\alpha}}]U_{ex}+i0^{+}$, $%
218: \{n_{\bar{\alpha}}\}\equiv n_{\bar{\alpha}}-[n_{\bar{\alpha}}]$, and $%
219: [n_{\alpha }]$ is the integer part of $n_{\alpha }$. $n_{\alpha }=n_{\alpha
220: \uparrow }+n_{\alpha \downarrow }$ is the total occupation number of
221: electron in the dot $\alpha $. After solving $\mathbf{g}_{\sigma
222: }^{r}(\epsilon )$ of the isolated DQDs, $G_{\alpha \alpha \sigma
223: }^{r}(\epsilon )$ and $G_{\alpha \alpha \sigma }^{<}(\epsilon )$ for the
224: whole system can be obtained from Dyson and Keldysh equations:\cite{addnote2}
225: \begin{eqnarray}
226: \mathbf{G}_{\sigma }^{r}(\epsilon )\equiv \left(
227: \begin{array}{ll}
228: G_{LL\sigma }^{r} & G_{LR\sigma }^{r} \\
229: G_{RL\sigma }^{r} & G_{RR\sigma }^{r}%
230: \end{array}%
231: \right) &=&\mathbf{g}_{\sigma }^{r}(\epsilon )+\mathbf{g}_{\sigma
232: }^{r}(\epsilon )\mathbf{\Sigma }_{\sigma }^{r}\mathbf{G}_{\sigma
233: }^{r}(\epsilon ), \\
234: \mathbf{G}_{\sigma }^{<}(\epsilon )\equiv \left(
235: \begin{array}{ll}
236: G_{LL\sigma }^{<} & G_{LR\sigma }^{<} \\
237: G_{RL\sigma }^{<} & G_{RR\sigma }^{<}%
238: \end{array}%
239: \right) &=&\mathbf{G}_{\sigma }^{r}(\epsilon )\mathbf{\Sigma }_{\sigma
240: }^{<}(\epsilon )\mathbf{G}_{\sigma }^{a}(\epsilon ).
241: \end{eqnarray}%
242: Here the bold face letters ($\mathbf{G}$, $\mathbf{g}$, and $\mathbf{\Sigma }
243: $) represent the $2\times 2$ matrix, and the self-energies $\mathbf{\Sigma }%
244: _{\sigma }^{r,<}(\epsilon )$ are:
245: \begin{eqnarray}
246: \mathbf{\Sigma }_{\sigma }^{r}(\epsilon ) &=&\left(
247: \begin{array}{cc}
248: -i\Gamma _{L}/2 & t_{c} \\
249: t_{c} & -i\Gamma _{R}/2%
250: \end{array}%
251: \right) , \\
252: \mathbf{\Sigma }_{\sigma }^{<}(\epsilon ) &=&\left(
253: \begin{array}{cc}
254: i\Gamma _{L}f_{L\sigma }(\epsilon ) & 0 \\
255: 0 & i\Gamma _{R}f_{R\sigma }(\epsilon )%
256: \end{array}%
257: \right) .
258: \end{eqnarray}%
259: Eqs. (3, 4, 5, and 6) can be solved self-consistently. The (charge) current
260: through the DQD is given by
261: \begin{equation*}
262: J=e(J_{L\uparrow }+J_{L\downarrow })=-e(J_{R\uparrow }+J_{R\downarrow }).
263: \end{equation*}
264:
265: Finally it is worth pointing out that the present problem can be solved by
266: other means, for example,\ the rate equation method.\cite{addnote4}
267:
268: \section{Spin-dependent Charge Stability Diagram and Charge Current}
269:
270: Before presenting numerical results, we emphasize that the spin bias we
271: apply to the DQDs device is a pure symmetric one without a (charge) bias,
272: i.e. $\mu _{L\uparrow }+\mu _{L\downarrow }=\mu _{R\uparrow }+\mu
273: _{R\downarrow }=0$.\cite{ref14} So if the spontaneously spin-polarized
274: occupations are not induced in the DQD, the charge current $J$ must be zero
275: because of the symmetric behaviors for the motion of spin-up electron and
276: the spin-down electron. For example, in the case of a single quantum dot
277: instead of DQDs applied by the pure spin bias, there is no spin polarization
278: in the dot and the current is always zero as the spin up-down symmetry is
279: retained. So, in the following, we first investigate the stability diagram
280: of spin polarization and the spin-dependent charge density in the DQD.
281:
282: Fig.2a and b present the spin polarizations $\Delta n_{\alpha }$ ($\Delta
283: n_{\alpha }\equiv n_{\alpha \uparrow }-n_{\alpha \downarrow }$) of the left
284: and right dots versus the levels $\epsilon _{L}$ and $\epsilon _{R}$, and
285: Fig. 2c presents the occupation number of electron $n_{L}+n_{R}/2$.\cite%
286: {note} It is found that these quantities are determined by the relative
287: energy levels of $\epsilon _{L}$ and $\epsilon _{R}.$ The spin polarization $%
288: \Delta n_{\alpha }$ is indeed non-zero and even quite large (i.e. near $\pm
289: 1 $) in some specific regions. Let us analyze the spin-dependent charge
290: stability diagram (see Fig.3a), which gives spin-dependent occupation
291: numbers of electron as a function of $\epsilon _{L}$ and $\epsilon _{R}$. If
292: without the spin bias ($V=0$), there are four domains $(0,1)$, $(1,1)$, $%
293: (0,2)$, and $(1,2)$ in the stability diagram (see the thin dashed curves in
294: Fig.3a), with $(n,m)$ representing $n$ and $m$ electrons in the left and
295: right dot. This type of charge stability diagram has been observed
296: experimentally,\cite{ref10,ref11} and is well established. While the spin
297: bias $V$ is turned on and the level $\epsilon _{L}$ or $\epsilon _{R}$
298: locates between $-V$ and $+V$, in addition of the four old spin-unpolarized
299: domains $(n,m)$ with a shift $V$ of their boundaries shift, there appears
300: four new spin-polarized domains, which are denoted by $(\uparrow ,1)$, $%
301: (0,\downarrow )$, $(1,\downarrow )$, and $(\uparrow ,2)$. The notation $%
302: (\uparrow ,1)$, for example, represents an electron of spin-up in the left
303: dot and a spin-unpolarized electron in the right dot.
304:
305: This spin-dependent charge stability diagram in Fig.3a can be obtained by
306: calculating the electrochemical potentials of the DQD or by analyzing the
307: level's position relative to the spin-dependent chemical potentials $\mu
308: _{\alpha \sigma }$. Consider the isolated DQDs device with $\Gamma
309: _{L}=\Gamma _{R}=t_{c}=0$. (i) The domain (0,1): when the equivalent level $%
310: \tilde{\epsilon}_{L}$ ($\tilde{\epsilon}_{L}\equiv \epsilon _{L}+U_{ex}$) of
311: the left dot is higher than $\mu _{L\uparrow }$ and $\mu _{L\downarrow },$
312: and the right-dot's level $\epsilon _{R}$ satisfies $\epsilon _{R}<\epsilon
313: _{L},\mu _{R\uparrow },\mu _{R\downarrow }<\epsilon _{R}+U_{in}$ (see
314: Fig.3b), the right dot is occupied by a spin-unpolarized electron and the
315: left dot is empty. (ii) The domain $(\uparrow ,1)$: while $\mu _{L\downarrow
316: }<\tilde{\epsilon}_{L}<\mu _{L\uparrow }$ and $\epsilon _{R}<\mu _{R\uparrow
317: },\mu _{R\downarrow }<\epsilon _{R}+U_{in}$ (see Fig.3c), a spin-up electron
318: occupies the left dot and a spin-unpolarized electron is in the right dot.
319: (iii) The domain $(0,\downarrow )$: if $\tilde{\epsilon}_{L}>\mu _{L\uparrow
320: },\mu _{L\downarrow }$ and $\mu _{R\uparrow }<\epsilon _{R}+U_{in}<\mu
321: _{R\downarrow }$ (see Fig.3d), the left dot is empty. For the right dot, a
322: spin-down electron occupies the level $\epsilon _{R}$ because of $\epsilon
323: _{R},\epsilon _{R}+U_{in}<\mu _{R\downarrow }$, then the spin-up level of
324: the right dot is pushed to $\epsilon _{R}+U_{in}$ which is over $\mu
325: _{R\uparrow }$, and so it is empty. Similarly, the other five domains can
326: also be obtained. In the case of the finite coupling case $\Gamma
327: _{L},\Gamma _{R},t_{c}\not=0$, the spin-polarized domains slightly extend to
328: the spin-unpolarized domains as illustrated in the thin dotted lines in
329: Fig.3a. Numerical results for the spin polarizations $\Delta n_{\alpha }$
330: (Fig.2a and b) and the occupation numbers of electrons $n_{L}+n_{R}/2$
331: (Fig.2c) are in a good agreement with the charge stability diagram in
332: Fig.3a. The eight domains, including four spin-unpolarized and four
333: spin-polarized domains, are clearly visible.
334:
335: In an alternative way, the stability diagram of Fig.3a can also be deduced
336: from the total energy of the DQD system and the electrochemical potentials.
337: When the isolated DQD is in the states of $\vec{N}=(N_{L\uparrow
338: },N_{L\downarrow },N_{R\uparrow },N_{R\downarrow }),$ where $N_{\alpha
339: \sigma }=0$ or $1$ is the index of the electron occupation number in the
340: intra-dot level $\alpha \sigma $, its total energy $E_{T}$ is
341: \begin{eqnarray}
342: E_{T}(\vec{N}) &=&N_{L}\epsilon _{L}+N_{R}\epsilon _{R}+N_{L}N_{R}U_{ex}
343: \notag \\
344: &&+(N_{L\uparrow }N_{L\downarrow }+N_{R\uparrow }N_{R\downarrow })U_{in},
345: \end{eqnarray}%
346: with $N_{\alpha }=N_{\alpha \uparrow }+N_{\alpha \downarrow }$. Consider the
347: fact that the occupation number in the intra-dot level $\alpha \sigma $ is
348: mainly effected by the lead (i.e. electron reservoir) $\alpha \sigma $. The
349: grand thermodynamic potential $\Omega $ at the zero temperature is
350: \begin{eqnarray}
351: \Omega (\vec{N}) &=&E_{T}(\vec{N})-N_{L\uparrow }\mu _{L\uparrow
352: }-N_{L\downarrow }\mu _{L\downarrow } \notag \\
353: &&-N_{R\uparrow }\mu _{R\uparrow }-N_{R\downarrow }\mu _{R\downarrow }.
354: \end{eqnarray}
355: In the present system, the electron occupation number can change
356: with the levels $\epsilon_L$ and $\epsilon_R$. This is a grand
357: canonical ensemble. Then the stablest state is one whose grand
358: thermodynamic potential $\Omega $ has the minimal values, and can be
359: found straightforwardly. For the sake of convenience and intuition,
360: we introduce the electrochemical potentials $\mu _{QD\alpha \sigma
361: }$, following Ref.\cite{ref10}. $\mu _{QD\alpha \sigma }$ of the
362: level $\alpha \sigma $ is well defined, for example,
363: \begin{equation}
364: \mu _{QDL\uparrow }(\vec{N})=E_{T}(\vec{N})-E_{T}(N_{L\uparrow
365: }-1,N_{L\downarrow },N_{R\uparrow },N_{R\downarrow }).
366: \end{equation}
367: Then the stablest states are the maximal values of $\vec{N}$ for which four $%
368: \mu _{QD\alpha \sigma }(\vec{N})$ are less than the corresponding chemical
369: potentials $\mu _{\alpha \sigma }$. If two states of $\vec{N}$, e.g. $\vec{N}%
370: =(0,0,1,0)$ and $(0,0,0,1),$ satisfy the above four equations, they are
371: assumed to have the same probability to exist. A detailed analysis of $\mu
372: _{QD\alpha \sigma }$ versus the parameters $\epsilon _{L}$ and $\epsilon
373: _{R} $ leads to establish the same charge stability diagram as shown in
374: Fig.3a. In fact, the electrochemical potentials $\mu _{QD\alpha \sigma }$
375: are equal to the equivalent levels in the preceding paragraph. For example,
376: \begin{eqnarray*}
377: \mu _{QDL\uparrow }(1,0,1,0) &=&\mu _{QDL\uparrow }(1,0,0,1)=\mu
378: _{QDL\downarrow }(0,1,1,0) \\
379: &=&\mu _{QDL\downarrow }(0,1,0,1)=\epsilon _{L}+U_{ex}=\tilde{\epsilon}_{L}.
380: \end{eqnarray*}
381: In particular, there are only four equivalent levels, which are less than
382: the numbers of $\mu _{QD\alpha \sigma }$. So it is convenient and intuitive
383: to use the equivalent levels to deduce the stability diagram.
384:
385: With the spin-polarized stability diagram in mind, we turn to calculate the
386: (charge) current $J$ induced by the spin bias. Fig.2d shows the current $J$
387: as a function of the levels $\epsilon _{L}$ and $\epsilon _{R}$. The current
388: becomes quite large when both the left and right dots are spin polarized in
389: the case of $-V<\tilde{\epsilon}_{L}=\epsilon _{R}+U_{in}<V$. The physical
390: origin of generation of the current has been explained in detail in the
391: introduction and as shown in Fig.1. We can establish a relation between the
392: charge current and the spin bias in the two leads. In this way, we can
393: detect the spin bias $V$ by measuring the current $J$. In the following we
394: calculate the current for various parameters. Fig.4a shows the current $J$
395: versus the spin bias $V$ for the inter-dot interaction $U_{ex}=5$. While $V=0
396: $, $J$ is zero exactly. With the increase of $V$ from zero, the current $J$
397: first increases, reaches at a maximum, and then drops. $J$ keeps a
398: relatively large value even if $V$ is comparable with the e-e interaction
399: energy $U_{in} $. The origin of the drop is that the spin-polarizations in
400: two dots decay while the current flows through the DQDs at the large $V$. In
401: the absence of the inter-dot e-e interaction $U_{ex}$, i.e. $U_{ex}=0$, the
402: current increases monotonously with the spin bias $V$ (see Fig.4b). In this
403: case the current $J$ and the spin bias $V$ have a one-to-one correspondence.
404: Therefore the spin bias $V$ can be deduced straightforward from the measured
405: current. Fig.4c shows the current $J$ as a function of the right-dot's level
406: $\epsilon _{R}$. When $\epsilon _{R}+U_{in}$ departs $\tilde{\epsilon}_{L}$
407: over a few $\Gamma _{\alpha }$ (e.g. $|\epsilon _{R}+U_{in}-\tilde{\epsilon}%
408: _{L}|>3\Gamma _{\alpha }$), $J$ becomes very small because the tunneling
409: process in Fig.1a suppresses quickly when $\epsilon _{R}+U_{in}$ is not in
410: alignment with $\tilde{\epsilon}_{L}$ . On the other hand, the tunneling
411: process in Fig.1a occurs frequently and $J$ becomes large when $\epsilon
412: _{R}+U_{in}$ is located near $\tilde{\epsilon}_{L}$. However, when $\epsilon
413: _{R}+U_{in}=\tilde{\epsilon}_{L}$, $J$ may drop slightly and a dip emerges
414: in the curve of $J$-$\epsilon _{R}$, because the spin-polarization $\Delta
415: n_{\alpha }$ is suppressed at the point. Fig.4d displays the current $J$ as
416: a function of temperature $T$. Here $J$ depends on the temperature $T$
417: slightly, and is quite large when $T<V$.
418:
419: \section{Charge bias in an open circuit}
420:
421: In the preceding section, we calculated the charge current through a DQD
422: induced by a pure spin bias. In an open circuit, the situation will be
423: changed. At the time that a spin bias is turned on, a charge current will
424: circulate. For an open circuit, the extra charge will accumulate in the two
425: leads until the system reaches at a balance. As a result an extra charge
426: bias $V_{e}$ instead of a charge current will be generated while the charge
427: current vanishes. In this case, combination of the the spin bias $V$ and the
428: induced charge bias $V_{e}$ will give the spin-dependent chemical potentials
429: $\mu _{\alpha \sigma }$ in the two leads
430: \begin{subequations}
431: \begin{eqnarray}
432: \mu _{L\uparrow } &=&+V+V_{e}, \\
433: \mu _{L\downarrow } &=&-V+V_{e}, \\
434: \mu _{R\uparrow } &=&-V-V_{e}, \\
435: \mu _{R\downarrow } &=&+V-V_{e}.
436: \end{eqnarray}%
437: The bias $V_{e}$ can be determined by the condition of
438: \end{subequations}
439: \begin{equation}
440: J=0
441: \end{equation}%
442: in equilibrium for an open circuit. Figs. 5a and 5b gives the bias $V_{e}$
443: and $V_{e}/V$ versus the spin bias $V$ in the presence and absence of the
444: intra-dot Coulomb interaction $U_{ex}$. $|V_{e}|$ and $|V_{e}/V|$ increase
445: monotonously with $V$ regardless of the value of $U_{ex}$. This is different
446: from the curve of $J$-$V$, in which $J$ drops down for a large $V$ while $%
447: U_{ex}\not=0$ (see Fig.4a). This illustrates that it is more efficient to
448: measure the induced bias $V_{e}$ than to measure the induced current $J$.
449: Fig.5c shows the bias $V_{e}$ as a function of the level $\epsilon _{R}$.
450: The bias $|V_{e}|$ always has a large value (e.g. $|V_{e}/V|>0.1$), even if $%
451: \epsilon _{R}+U_{in}$ is far away from $\tilde{\epsilon}_{L}$. Notice that
452: the current $J$ is relatively small when $|\epsilon _{R}+U_{in}-\tilde{%
453: \epsilon}_{L}|>3\Gamma _{\alpha }$ (see Fig.4c). The transmission
454: coefficient (or the conductance) is also very small in this region.
455: Correspondingly, $V_{e}$ in an open circuit is still large.
456: Therefore the induced bias $V_{e}$ can be measured in an more
457: extensive region. Fig.5d gives the temperature $T$ dependence of the
458: bias $V_{e}$, which is almost independent of the temperature $T$.
459: Finally, we emphasize that $|V_{e}/V|$ is usually larger than $0.1$
460: regardless of the values of the parameters $V$, $\epsilon _{L}$,
461: $\epsilon _{R}$, $T$, etc. In the current technology, the bias in
462: the order of $0.1nV$ is measurable in experiment.\cite{add21}
463: Therefore, if the spin bias $V$, i.e. the difference of the spin-up
464: and spin-down chemical potentials $(\mu _{L\uparrow }-\mu
465: _{L\downarrow })/e$, reaches to $1nV$, the induced bias in the
466: present calculation is large enough to be measured in experiment.
467:
468: \section{conclusions}
469:
470: In summary, we investigated the electron transport driven by a spin
471: bias or pure spin current through a non-magnetic DQD. Except for the
472: spin-unpolarized domains, several spin-polarized domains are found
473: in the stability diagram with respect to the energy levels of two
474: quantum dots. When both of the left and right dots are spin
475: polarization, a large charge current $J$ can be induced by applying
476: of the pure spin bias. In particular, in an open circuit, the charge
477: bias is induced to balance the spin bias, and is measurable in an
478: extensive range of the parameters. Physically, a pure spin bias may
479: drive electrons with different spin in opposite direction. If the
480: system possesses the left-right symmetry or parity and does not
481: break the time reversal symmetry it will circulate a pure spin
482: current (or spin accumulation in an open circuit). When the energy
483: levels in the two dots are not equal, the left-right symmetry or
484: parity of the system is broken. A spin bias and strong Coulomb
485: interaction can produce two spin polarized states in the two dots as
486: we discussed in the spin-polarized charge stability diagram. As a
487: result the currents with different spins in opposite directions will
488: not be equal any more. Consequently this pure spin bias generates a
489: charge current through the DQD. This property may provide a
490: practical approach to detect the spin bias in DQD by measuring the
491: charge bias or charge current.
492:
493: \section*{Acknowledgments}
494:
495: We acknowledge the financial support from NSF-China under Grant
496: Nos. 10525418, 10734110, and 60776060 (Q.F.S.), and the Research
497: Grant Council of Hong Kong under Grant No.: HKU 7041/07P (S.Q.S.).
498: Q.F.S. would like to thank Dr. W. Long for many helpful
499: discussions.
500:
501: \begin{references}
502:
503: \bibitem{ref1}
504: S. A. Wolf, D. D. Awschalom, R. A. Buhrman, J. M. Daughton, S. von
505: Moln$\acute{a}$r, M. L. Roukes, A. Y. Chtchelkanova, and D. M.
506: Treger, Science {\bf 294}, 1488 (2001); Gary A. Prinz, Science {\bf
507: 282}, 1660 (1998).
508:
509: \bibitem{ref2}
510: I. Zutic, J. Fabian, and S. Das Sarma, Rev. Mod. Phys. {\bf 76}, 323
511: (2004).
512:
513: \bibitem{Awschalom}
514: D. D. Awschalom and M. E. Flatte, Nature Phys. {\bf 3}, 153 (2007).
515:
516: \bibitem{ref3}
517: R. D. R .Bhat and J. E. Sipe, Phys. Rev. Lett. {\bf 85}, 5432
518: (2000); Q.-F. Sun, H. Guo, and J. Wang, {\sl ibid}. {\bf 90},
519: 258301 (2003); P. Sharma and P. W. Brouwer, {\sl ibid}. {\bf 91},
520: 166801 (2003); Susan K. Watson, R. M. Potok, C. M. Marcus, and V.
521: Umansky, {\sl ibid}. {\bf 91}, 258301 (2003); W. Long, Q.-F. Sun,
522: Hong Guo, and J. Wang, Appl. Phys. Lett. {\bf 83}, 1397 (2003).
523:
524: \bibitem{ref4}
525: J. H$\ddot{u}$bner, W. W. R$\ddot{u}$hle, M. Klude, D. Hommel, R. D.
526: R. Bhat, J. E. Sipe, and H. M. van Driel, Phys. Rev. Lett. {\bf 90},
527: 216601 (2003); M. J. Stevens, Arthur L. Smirl, R. D. R. Bhat, Ali
528: Najmaie, J. E. Sipe, and H. M. van Driel, {\sl ibid}. {\bf 90},
529: 136603 (2003).
530:
531: \bibitem{cui}
532: X. D. Cui, S.-Q. Shen, J. Li, Y. Ji, W.K. Ge, and F.-C. Zhang,
533: Appl. Phys. Lett. {\bf 90}, 242115 (2007); J. Li, X. Dai, S.-Q.
534: Shen, and F.-C. Zhang, {\sl ibid}. {\bf 88}, 162105 (2006).
535:
536: \bibitem{ref5}
537: S. O. Valenzuela and M. Tinkham, Nature (London) {\bf 442}, 176
538: (2006).
539:
540: \bibitem{Kimura}
541: T. Kimura, Y. Otani, T. Sato, S. Takahashi and S. Maekawa, Phys.
542: Rev. Lett. 98, 156601 (2007); E. Saitoh, M. Ueda, H. Miyajima, and
543: G. Tatara, Appl. Phys. Lett. {\bf 88}, 182509 (2006).
544:
545: \bibitem{ref6}
546: Y. K. Kato, R. C. Myers, A. C. Gossard, and D. D. Awschalom, Science
547: {\bf 306}, 1910 (2004); V. Sih, R. C. Myers, Y. K. Kato, W. H. Lau,
548: A. C. Gossard, and D. D. Awschalom, Nature Phys. {\bf 1}, 31 (2005);
549: V. Sih, W. H. Lau, R. C. Myers, V. R. Horowitz, A. C. Gossard, and
550: D. D. Awschalom, Phys. Rev. Lett. {\bf 97}, 096605 (2006).
551:
552: \bibitem{ref7}
553: J. Wunderlich, B. Kaestner, J. Sinova, and T. Jungwirth, Phys.
554: Rev. Lett. {\bf 94}, 047204 (2005).
555:
556: \bibitem{ref8}
557: P. Mohanty, G. Zolfagharkhani, S. Kettemann, and P. Fulde, Phys.
558: Rev. B {\bf 70}, 195301 (2004); Tsung-Wei Chen, Chih-Meng Huang, and
559: G. Y. Guo, {\sl ibid}. {\bf 73}, 235309 (2006).
560:
561: \bibitem{addref1}
562: F. Meier and D. Loss, Phys. Rev. Lett. {\bf 90}, 167204 (2003); F.
563: Sch$\ddot{u}$tz, M. Kollar, and P. Kopietz, {\sl ibid}. {\bf 91},
564: 017205 (2003).
565:
566: \bibitem{ref9}
567: Q.-F. Sun, H. Guo, and J. Wang, Phys. Rev. B {\bf 69}, 054409
568: (2004); Q.-F. Sun and X.C. Xie, {\sl ibid}. {\bf 72}, 245305 (2005).
569:
570: \bibitem{add1}
571: G. Bergmann, Phys. Rev. B {\bf 63}, 193101 (2001); A. G.
572: Mal'shukov, C.S. Tang, C.S. Chu, and K.A. Chao, Phys. Rev. B {\bf
573: 68}, 233307 (2003); S. I. Erlingsson and D. Loss, Phys. Rev. B
574: {\bf 72}, 121310(R) (2005).
575:
576: \bibitem{addnote1}
577: For the persistent spin current in the equilibrium system, the
578: spin bias is always zero, which is very similar the supercurrent
579: in a superconducting state. So the method in the present paper
580: cannot be applied to detect the persistent spin current. The
581: potential measurement methods for this type of spin current is
582: referred to Q.-F. Sun, X.C. Xie, and J. Wang, Phys. Rev. Lett.
583: {\bf 98}, 196801 (2007); Phys. Rev. B 77, 035327 (2008), and
584: references therein.
585:
586: \bibitem{ref10}
587: W. G. van der Wiel, S. De Franceschi, J. M. Elzerman, T. Fujisawa,
588: S. Tarucha, and L. P. Kouwenhoven, Rev. Mod. Phys. {\bf 75}, {\bf 1}
589: (2003).
590:
591: \bibitem{ref11}
592: R. Hanson, L. P. Kouwenhoven, J. R. Petta, S. Tarucha, and L. M. K.
593: Vandersypen, Rev. Mod. Phys. {\bf 79}, 1217 (2007).
594:
595: \bibitem{ref12}
596: D. Loss and D .P. DiVincenzo, Phys. Rev. A {\bf 57}, 120 (1998);
597: A. Imamoglu, D.D. Awschalom, G. Burkard, D.P. DiVincenzo, D. Loss,
598: M. Sherwin, and A. Small, Phys. Rev. Lett. {\bf 83}, 4204 (1999);
599: B. Burkard and D. Loss, {\sl Semiconductor Spintronics and Quantum
600: Computation}, edited by D.D. Awschalom {\sl et al.}
601: (Springer-Verlag, Berlin, 2002), Chap. 8.
602:
603: \bibitem{ref13}
604: K. Ono and S. Tarucha, Phys. Rev. Lett. {\bf 92}, 256803
605: (2004).
606:
607: \bibitem{ref14}
608: D.-K. Wang, Q.-F. Sun, and H. Guo, Phys. Rev. B {\bf 69}, 205312
609: (2004).
610:
611: \bibitem{ref15}
612: Y. Meir and N. S. Wingreen, Phys. Rev. Lett. {\bf 68}, 2512 (1992).
613:
614:
615: \bibitem{note21}
616: Here we took the approximation: $\langle \hat{n}_{L\sigma}
617: \hat{n}_{R\sigma'} \rangle = \langle \hat{n}_{L\sigma} \rangle
618: \langle \hat{n}_{R\sigma'} \rangle = n_{L\sigma}n_{R\sigma'}$, and
619: $\langle \hat{n}_{\alpha\uparrow} \hat{n}_{\alpha\downarrow}
620: \rangle =0$ while $0<n_{\alpha}<1$ and $\langle
621: \hat{n}_{\alpha\uparrow} \hat{n}_{\alpha\downarrow} \rangle
622: =n_{\alpha} -1$ while $1<n_{\alpha}<2$, where
623: $\hat{n}_{\alpha\sigma} \equiv
624: d^{\dagger}_{\alpha\sigma}d_{\alpha\sigma}$. Because of
625: $V<U_{in}$, the fluctuation of the occupation number is
626: less than one in each dot, and the approximation should be
627: reasonable.
628:
629: \bibitem{addnote2}
630: This system with the e-e interactions in the
631: non-equilibrium case (e.g. with the non-zero spin bias) is not exactly solvable.
632: One has to introduce some approximations. In
633: the equations (5) and (6), we have neglected the higher-order of
634: self-energy correction which originate from the combination of the e-e
635: interaction and the hopping terms. This approximation is much
636: better than the Hartree-Fork approximation. When the hopping
637: coefficients ($t_{\alpha}$ and $t_c$) are weak or when the system
638: is not in the Kondo regime, this method is expected to work very well, e.g. see
639: A.~Groshev, T.~Ivanov, and V.~Valtchinov, Phys. Rev. Lett.
640: \textbf{66}, 1082 (1991); Q.-F. Sun and X.C. Xie, Phys. Rev. B
641: {\bf 73}, 235301 (2006).
642:
643: \bibitem{addnote4}
644: S.A. Gurvitz and Ya.S. Prager, Phys. Rev. B {\bf 53},
645: 15932 (1996); S.A. Gurvitz, Phys. Rev. B {\bf 57}, 6602 (1998).
646:
647: \bibitem{note}
648: The occupation numbers of electron, in particular $n_L +n_R/2$, can be
649: measured by using the charge sensing techniques, i.e., using a
650: quantum point contact (QPC) nearby the DQDs device to measure the
651: conductance of the QPC. For the detail, see Refs.\cite{ref10,ref11}.
652:
653:
654:
655: \bibitem{add21}
656: H. Safar, P. L. Gammel, D. A. Huse, and D. J. Bishop, J. P. Rice and
657: D. M. Ginsberg, Phys. Rev. Lett. {\bf 69}, 824 (1992).
658:
659:
660: \end{references}
661:
662: \begin{figure}[tbph]
663: %\centering \includegraphics[width=0.45\textwidth]{fig1.eps} \newline
664: \caption{ (color online) (a) ((b)) The schematic plots illustrate an spin-up
665: (spin-down) electron tunneling from the left (right) to the right (left)
666: lead. }
667: \label{fig::fig1}
668: \end{figure}
669:
670: \begin{figure}[tbph]
671: %\centering \includegraphics[width=0.45\textwidth]{fig2.eps} \newline
672: \caption{(color online) The spin polarization $\Delta n_{L}$ at the left dot
673: (a), $\Delta n_{R}$ at the right dot (b), $n_{L}+n_{R}/2$ (c), and the
674: current $J$ (d) as a function of the energy levels $\protect\epsilon _{L}$
675: and $\protect\epsilon _{R}$ in the two quantum dots. The parameters are: $%
676: \Gamma _{L}=\Gamma _{R}=0.3$, $t_{c}=T=0.1$, $U_{in}=20$, $U_{ex}=5$, and $%
677: V=1$. }
678: \label{fig::fig2}
679: \end{figure}
680:
681: \begin{figure}[tbph]
682: %\centering \includegraphics[width=0.45\textwidth]{fig3.eps} \newline
683: \caption{ (color online) (a) Schematic stability diagram of the DQD under
684: the finite spin bias $V$. The thin dashed lines are the stability diagram
685: while $V=0$. (b), (c), and (d) are location schematics of energy levels in
686: the $(0,1)$ , $(\uparrow ,1)$, and $(0,\downarrow )$ domains, respectively. }
687: \label{fig::fig3}
688: \end{figure}
689:
690: \begin{figure}[tbph]
691: %\centering \includegraphics[width=0.45\textwidth]{fig4.eps} \newline
692: \caption{The current $J$ vs. the spin bias $V$ for two inter-dot
693: interactions $U_{ex}=5$ (a) and $U_{ex}=0$ (b). (c) The current $J$ vs. the
694: level $\protect\epsilon _{R}$ and (d) the current $J$ vs. the temperature $T$%
695: . The solid, dashed, and dotted curves are for levels at $\tilde{\protect%
696: \epsilon}_{L}=0$ and $\protect\epsilon _{R}=-20$, $\tilde{\protect\epsilon}%
697: _{L}=0.5$ and $\protect\epsilon _{R}=-19.5$, and $\tilde{\protect\epsilon}%
698: _{L}=1$ and $\protect\epsilon _{R}=-19$, respectively. The other parameters
699: are the same as in Fig.2. }
700: \label{fig::fig4}
701: \end{figure}
702:
703: \begin{figure}[tbph]
704: %\centering \includegraphics[width=0.45\textwidth]{fig5.eps} \newline
705: \caption{The induced charge bias $V_{e}$ (thin curves) and $V_{e}/V$ (thick
706: curves) vs. the spin bias $V$ for inter-dot interaction $U_{ex}=5$ (a) and $%
707: U_{ex}=0$ (b). (c) $V_{e}$ (i.e. $V_{e}/V$) vs. the level $\protect\epsilon %
708: _{R}$ and (d) $V_{e}$ (i.e. $V_{e}/V$) vs. the temperature $T$. The solid,
709: dashed, and dotted curves are for levels at $\tilde{\protect\epsilon}_{L}=0$
710: and $\protect\epsilon _{R}=-20$, $\tilde{\protect\epsilon}_{L}=0.5$ and $%
711: \protect\epsilon _{R}=-19.5$, and $\tilde{\protect\epsilon}_{L}=1$ and $%
712: \protect\epsilon _{R}=-19$, respectively. The other parameters are the same
713: as in Fig.2 }
714: \label{fig::fig5}
715: \end{figure}
716:
717: \end{document}
718: