1: \documentclass[twocolumn]{jpsj2}
2: % \documentclass[preprint]{jpsj2}
3: \usepackage{graphicx}
4: \renewcommand{\baselinestretch}{0.833}
5:
6: \def\runtitle{Current-Induced Entanglement of Nuclear Spins in Quantum Dots}
7: \def\runauthor{M.\ {\sc Eto}, T.\ {\sc Ashiwa} and M.\ {\sc Murata}}
8:
9: \title{%
10: Current-Induced Entanglement of Nuclear Spins in Quantum Dots
11: }
12:
13: \author{%
14: Mikio {\sc Eto},\thanks{E-mail address: eto@rk.phys.keio.ac.jp}
15: Takashi {\sc Ashiwa} and Mikio {\sc Murata}
16: }
17:
18: \inst{%
19: Faculty of Science and Technology, Keio University, \\
20: 3-14-1 Hiyoshi, Kohoku-ku, Yokohama 223-8522, Japan
21: }
22:
23: \recdate{August 5, 2003}
24:
25: \abst{%
26: We propose an entanglement mechanism of nuclear spins in quantum dots
27: driven by the electric current accompanied by the spin flip.
28: This situation is relevant to a leakage current in spin-blocked regions
29: where electrons cannot be transported unless their spins are flipped.
30: The current gradually increases the components of larger total spin of nuclei.
31: This correlation among the nuclear spins markedly enhances
32: the spin-flip rate of electrons and hence the leakage current.
33: The enhancement of the current is observable when the residence time
34: of electrons in the quantum dots is shorter than the dephasing time
35: $T_2^*$ of nuclear spins.
36: }
37:
38: \kword{%
39: quantum dot, entanglement, nuclear spins, hyperfine interaction,
40: spin blockade, dephasing, superradiance, Dicke effect}
41:
42: \begin{document}
43: \sloppy
44: \maketitle
45:
46: The spin relaxation of electrons in quantum dots is an important issue.
47: The relaxation time must be sufficiently long for the
48: implementation of quantum computing devices utilizing electron spins
49: in quantum dots.\cite{Loss} The time has been examined experimentally by
50: optical\cite{opt1,opt2} and transport measurements.\cite{Fujisawa,Ono}
51: Mechanisms of the spin relaxation have been investigated theoretically
52: by, {\it e.g.}, hyperfine interaction,\cite{Sigurdur,
53: Sigurdur2,Merkulov,Lyanda-Geller,Khaetskii,Schliemann}
54: spin-orbit interaction\cite{Khaetskii1,Khaetskii2} and
55: higher-order tunneling processes (cotunneling).\cite{Averin,Eto}
56:
57: In this letter, we study the unique properties of nuclear spin states
58: in quantum dots, which stem from the spin relaxation of electrons
59: through hyperfine interaction.
60: In quantum dots, an electron occupying a particular level
61: couples with numerous nuclear spins simultaneously through the interaction.
62: Hence a spin flip of the electron results in a correlation among the
63: nuclear spins. The correlation is an entanglement: we do not know which
64: of the nuclear spins is flipped.
65: We show that (i) the electric current accompanied by the spin flip
66: significantly increases the correlation among nuclear spins, and
67: (ii) the correlated state of nuclear spins markedly
68: enhances the spin-flip rate of electrons.
69: The direct dipole-dipole interaction between the nuclear spins
70: is much smaller than the hyperfine interaction\cite{Merkulov}
71: and is thus disregarded. Its influence is discussed later.
72:
73: This situation is relevant to a leakage current in the spin-blocked region
74: where electrons cannot be transported unless the spins are flipped. In weakly
75: coupled double quantum dots, Ono {\it et al}.\ have observed a current
76: suppression when two electrons occupy the lowest energy level in
77: each dot with parallel spins;\cite{Ono} an electron tunneling from one dot
78: to the other is forbidden by the Pauli exclusion principle (see Fig.\ 4).
79: In this region, we predict that a leakage current
80: accompanied by the spin flip in one of the dots is significantly enhanced by
81: the nuclear spin entanglement.
82: If our theory is verified, it would be the first observation of entanglement
83: in solid-state devices.
84:
85: Our mechanism is analogous to the current-induced dynamic nuclear
86: polarization (DNP) in quantum Hall systems.\cite{DNP1,DNP2,DNP3,Machida}
87: The DNP is created by electron scattering between spin-polarized edge
88: states, accompanied by the spin flip. The DNP influences the electronic
89: state and, in consequence, leads to a hysteresis in the longitudinal
90: resistance.\cite{Kronmuller,Smet} Note that nuclear spins are entangled but
91: not polarized in a specific direction in our case.
92:
93:
94: % {\it Model}:
95: We examine a simple model for the spin-blocked quantum dots and
96: discuss its relevance to a realistic situation later.
97: Our model is as follows.
98: A quantum dot is connected to external leads, $L$, $R$, through tunnel
99: barriers.
100: The Coulomb blockade restricts the number of electrons in the dot to be
101: $N_{\rm el}$ or $N_{\rm el}+1$. $N_{\rm el}$ electrons form a background
102: of a spin singlet.
103: From lead $L$ on the source side, an extra electron tunnels into the dot
104: and occupies a single level with an envelope wavefunction
105: $\psi({\mib r})$. The spin of the electron is either $| \uparrow \rangle$
106: ($S_z=1/2$) or $| \downarrow \rangle$ ($S_z=-1/2$) with equal probability
107: (we assume an easy-axis of electron spins in the $z$ direction, {\it e.g}.,
108: in a magnetic field). The electron stays for a long time by the spin
109: blockade, interacting with $N$ nuclear spins,
110: ${\mib I}_k$ (located at ${\mib r}_k$; $k=1,\cdots,N$),
111: through the hyperfine contact interaction.
112: After the spin is flipped, the electron tunnels out to lead $R$ on the
113: drain side, and the next electron is immediately injected from lead $L$.
114:
115: The hyperfine interaction is expressed as
116: \begin{equation}
117: H_{\rm hf} = 2 \sum_{k=1}^N \alpha_k {\mib S} \cdot {\mib I}_k,
118: \label{eq:Hamil0}
119: \end{equation}
120: where $\alpha_k \propto |\psi({\mib r}_k)|^2$.
121: $N \sim 10^5$ in GaAs quantum dots.
122: We consider nuclear spins of $1/2$ for simplicity. We assume that
123: the spin-flip rate from $| \downarrow \rangle$ to $| \uparrow \rangle$
124: ($| \uparrow \rangle$ to $| \downarrow \rangle$) is given by the perturbation
125: of $H_{\rm hf}$,
126: \begin{equation}
127: \Gamma=A |\langle \Psi_{\rm fin}; \uparrow (\downarrow)
128: |H_{\rm hf}| \Psi_{\rm init}; \downarrow (\uparrow) \rangle |^2,
129: \label{eq:Gamma}
130: \end{equation}
131: where $\Psi_{\rm init}$ and $\Psi_{\rm fin}$ are the initial and
132: final states of nuclear spins, respectively.\cite{com0}
133: We first study the case of homogeneous coupling constants, $\alpha_k = \alpha$.
134: This is a good approximation for the majority of nuclear
135: spins since the distance between neighboring nuclei is much smaller than
136: the size of the quantum dot.
137: Then the Hamiltonian (\ref{eq:Hamil0}) is reduced to
138: \begin{equation}
139: H_{\rm hf} = 2 \alpha {\mib S} \cdot {\mib I},
140: \label{eq:Hamil}
141: \end{equation}
142: where ${\mib I}=\sum_{k=1}^N {\mib I}_k$ is the total spin of nuclei.
143: A generic case with inhomogeneous hyperfine coupling is examined later.
144:
145: % {\it Basic idea}:
146: The Hamiltonian (\ref{eq:Hamil}) [and also (\ref{eq:Hamil0})]
147: indicates that $N$ nuclear spins interact
148: with a common ``field'' of an electron spin although there is no direct
149: interaction between them (dipole-dipole interaction between nuclear spins
150: is weak and neglected). This field results in an entanglement among
151: nuclear spins. To illustrate this, let us consider the simplest case of $N=2$
152: and begin with a polarized state of nuclear spins
153: $| I_{z}=1/2\rangle_1 | I_{z}=1/2 \rangle_2$. An electron with spin
154: $| \downarrow \rangle$ tunnels into the dot and is spin-flipped.
155: Then the state of nuclear spins becomes
156: $(| -1/2 \rangle_1 | 1/2 \rangle_2 + | 1/2 \rangle_1 | -1/2\rangle_2 )
157: /\sqrt{2} =| J=1,M=0 \rangle$, where $J$ and $M$ are the total spin and its
158: $z$ component, respectively. This is an entangled state; we do not know
159: which of the nuclear spins is flipped.
160: After the electron tunnels out of the dot, the next electron is injected with
161: $| \downarrow \rangle$ (or $| \uparrow \rangle$) and interacts with this
162: state of nuclear spins.
163: The spin-flip rate for the second electron is given by $2A\alpha^2$.
164: This value is twice the rate in the case of non-entangled states
165: of nuclear spins,
166: $|1/2\rangle_1 | -1/2 \rangle_2$ or $|-1/2\rangle_1 | 1/2 \rangle_2$.
167: In general, the capability of state $|J,M \rangle$ to
168: flip an electron spin is $A\alpha^2 (J \pm M)(J \mp M +1)$.
169:
170: % {\it In presence of easy-axis}:
171: Now we formulate the case of $N$ nuclear spins.
172: We assume that the initial state of the nuclear spins is
173: \begin{eqnarray}
174: \Psi^{(0)} & = & \sum_{I_{z,1}=\pm 1/2} \cdots \sum_{I_{z,N}=\pm 1/2}
175: C(I_{z,1},\cdots,I_{z,N})
176: \nonumber \\
177: & & \times |I_{z,1} \rangle_1 |I_{z,2} \rangle_2 \cdots |I_{z,N} \rangle_N,
178: \label{eq:Psi00}
179: \end{eqnarray}
180: where $\{ C(I_{z,1},\cdots,I_{z,N}) \}$ are randomly
181: distributed.\cite{com1}
182: By the transformation of the basis set into the eigenstates of the total
183: spin $J$ and its $z$ component $M$ of all the nuclei
184: ($J \le N/2$, $-J \le M \le J$), eq.\ (\ref{eq:Psi00}) is rewritten as
185: \begin{equation}
186: \Psi^{(0)}=\sum_{J,M,\lambda} C_{J,M,\lambda}^{(0)} | J,M,\lambda \rangle,
187: \label{eq:Psi0}
188: \end{equation}
189: with random coefficients $\{ C_{J,M,\lambda}^{(0)} \}$
190: ($\sum_{J,M,\lambda} |C_{J,M,\lambda}^{(0)}|^2=1$). Index
191: $\lambda$ distinguishes states with the same $J$ and $M$.
192: The number of such states is
193: \begin{equation}
194: K(N,J)=(2J+1)\frac{N!}{(\frac{N+2J+2}{2})! (\frac{N-2J}{2})!}.
195: \end{equation}
196:
197: The first electron with $| \downarrow \rangle$ (or $| \uparrow \rangle$)
198: enters the dot and interacts with $\Psi^{(0)}$ of nuclear spins. The
199: spin-flip rate is $\Gamma^{(0)}=A \alpha^2 F^{(0)}$, where
200: \begin{equation}
201: F^{(0)}=\sum_{J,M,\lambda} |C_{J,M,\lambda}^{(0)}|^2(J \pm M)(J \mp M+1).
202: \end{equation}
203: When $N \gg 1$, $|C_{J,M,\lambda}^{(0)}|^2$ can be replaced by
204: $1/2^N$ (law of large number; fluctuation of $1/\sqrt{2^N}$ is neglected).
205: Then $F^{(0)}=(1/2^N)\sum_{J,M} K(N,J) (J \pm M)(J \mp M+1) \approx N/2$.
206: Note that $\Gamma^{(0)}=(N/2) A \alpha^2$ is identical to the spin-flip rate
207: which would be evaluated under the assumption that one of the nuclear spins
208: is flipped with the electron spin.\cite{com3}
209:
210: %%%%%%%%%%%%%%
211: \begin{figure}[hbt]
212: \begin{center}
213: \includegraphics[width=7.5cm]{fig1a.eps} \\
214: \includegraphics[width=7.5cm]{fig1b.eps}
215: \end{center}
216: \caption{Distribution of (a) total spin $J$ and (b) its $z$
217: component in the state of nuclear spins.
218: (a) $p(J)=\sum_{M,\lambda} |C_{J,M,\lambda}^{(n)}|^2$, and
219: (b) $p(M)=\sum_{J,\lambda} |C_{J,M,\lambda}^{(n)}|^2$, where
220: $n$ is the number of transported electrons accompanied by the spin flip.
221: The number of nuclear spins is $N=100$. In eq.\ (\ref{eq:Psin}),
222: we take the geometrical mean between upper and lower signs.}
223: \end{figure}
224: %%%%%%%%%%%%%%
225:
226: After the spin flip, the electron tunnels off the dot.
227: The state of nuclear spins becomes
228: \begin{eqnarray}
229: \Psi^{(1)} = \frac{1}{\sqrt{F^{(0)}}}\sum_{J,M,\lambda} C_{J,M,\lambda}^{(0)}
230: \sqrt{(J \pm M)(J \mp M+1)}
231: \nonumber \\
232: \times | J,M \mp 1,\lambda \rangle.
233: \end{eqnarray}
234: $\Psi^{(1)}$ includes more components of larger $J$ and smaller $|M|$.
235: The correlation among the nuclear spins increases
236: each time an electron is injected, spin-flipped, and ejected out of the dot.
237: After $n$ such events, the state of nuclear spins is
238: \begin{eqnarray}
239: \Psi^{(n)} & = & \sum_{J,M,\lambda} C_{J,M,\lambda}^{(n)}
240: | J,M,\lambda \rangle, \\
241: \label{eq:Psin0}
242: C_{J,M \mp 1,\lambda}^{(n)} & = &
243: \sqrt{\frac{(J \pm M)(J \mp M+1)}{F^{(n-1)}}} C_{J,M,\lambda}^{(n-1)},
244: \label{eq:Psin}
245: \\
246: F^{(n)} & = &
247: \sum_{J,M,\lambda} |C_{J,M,\lambda}^{(n)}|^2 (J \pm M)(J \mp M+1).
248: \label{eq:Psin2}
249: \end{eqnarray}
250: $F^{(n)}$ is expressed as $F^{(n)} = f^{(n)} / f^{(n-1)}$ with
251: \begin{equation}
252: f^{(n)} = \frac{1}{2^N}\sum_{J,M} K(N,J)
253: \left[ (J \pm M)(J \mp M+1) \right]^n.
254: \label{eq:Fn}
255: \end{equation}
256: Figure 1 shows the distribution of (a) the total spin $J$ and
257: (b) its $z$ component $M$. With increasing $n$, the weight of
258: larger $J$ and smaller $|M|$ increases. This means that
259: the total nuclear spins are developed in the plane perpendicular to
260: the easy-axis of electron spins.
261:
262: The correlation among the nuclear spins enhances the spin-flip rate.
263: The rate for the $(n+1)$th electron is given by
264: $\Gamma^{(n)}=A\alpha^2 F^{(n)}$. Using eq.\ (\ref{eq:Fn}), we find that
265: $F^{(n)} \approx (N/2)n$ for $1 \ll n \ll N/2$, and
266: $F^{(n)} \approx (N/2)^2$ for $N/2 \ll n$. Therefore the spin-flip
267: rate increases with $n$ linearly
268: ($\Gamma^{(n)} \approx n \Gamma^{(0)}$) and finally saturates
269: ($\Gamma^{(n)} \approx (N/2) \Gamma^{(0)}$), as shown in Fig.\ 2(a).
270:
271: %%%%%%%%%%%%%%
272: \begin{figure}[hbt]
273: \begin{center}
274: \includegraphics[width=7cm]{fig2a.eps} \\
275: \includegraphics[width=7cm]{fig2b.eps}
276: \end{center}
277: \caption{(a) Spin-flip rate $\Gamma^{(n)}$
278: as a function of $n$ (number of transported electrons accompanied by
279: spin flip) and (b) electric current $I(t)$ as a function of time $t$,
280: with homogeneous hyperfine coupling.
281: The number of nuclear spins is $N=100$.
282: We take the geometrical mean between upper and lower signs in
283: eq.\ (\ref{eq:Fn}).}
284: \end{figure}
285: %%%%%%%%%%%%%%
286:
287: We evaluate the electric current $I(t)$.
288: The time interval between $n$th and $(n+1)$th electron transports is
289: $\Delta t^{(n)} \approx 1/\Gamma^{(n)}$. The current as a function of $t$,
290: $I(t)=e\Gamma^{(n)}$ at $t=\sum_{j=0}^{n} \Delta t^{(j)}$,
291: is shown in Fig.\ 2(b). The approximate form of
292: $\Gamma^{(n)}$ yields an asymptotic form of the current,
293: \begin{equation}
294: \left\{
295: \begin{array}{llll}
296: I(t) & \approx & e\Gamma^{(0)} e^{\Gamma^{(0)} t} & \hspace{.5cm}
297: (t \ll t_{\rm sat}) \\
298: I(t) & \approx & e(N/2)\Gamma^{(0)} & \hspace{.5cm} (t_{\rm sat} \ll t).
299: \end{array}
300: \right.
301: \label{eq:current}
302: \end{equation}
303: The current grows significantly with time and finally saturates.
304: The saturation time is given by
305: $t_{\rm sat}=\ln(N/2)/\Gamma^{(0)}$, where $\Gamma^{(0)}$ is the spin-flip
306: rate of an electron with the uncorrelated state of nuclear spins.
307:
308:
309: % {\it Inhomogeneous hyperfine coupling}:
310: Up to now, we have examined the case of
311: homogeneous hyperfine coupling in the quantum dot,
312: $\alpha_k=\alpha$. Generally, the coupling constant
313: $\alpha_k$ depends on the position of the nucleus due to the spatial
314: variation of $\psi({\mib r})$. With this inhomogeneous coupling, (i)
315: the total spin $J$ is no longer a good quantum number, and
316: (ii) the correlated state of nuclear spins is dephased.
317: During the stay of an electron in the dot
318: [$\tau \sim 1/\Gamma^{(0)}=2/(A\alpha^2 N)$],
319: each term in eq.\ (\ref{eq:Psi00}) gains a phase factor
320: $e^{i\omega \tau}$ with $\omega=\pm \sum_k \alpha_k I_{z,k} /\hbar$.
321: This factor diminishes the correlation if the residence time $\tau$ is
322: sufficiently long.
323:
324: To examine these effects,
325: we perform numerical calculations in the case of
326: $\alpha_k \propto \exp{-\left[ (k-1) / x_0 \right]^2}$ ($k=1,2,\cdots,N$;
327: $\frac{1}{N}\sum_k\alpha_k=\alpha$).
328: The number of nuclear spins is $N=14$.
329: Electrons with spin $| \uparrow \rangle$ and $| \downarrow \rangle$ are
330: injected into the dot alternately.
331: For the initial state of nuclear spins, $\Psi^{(0)}$,
332: the random average is taken over the coefficients
333: $\{ C(I_{z,1},\cdots,I_{z,N}) \}$ in eq.\ (\ref{eq:Psi00}).
334:
335: Figure 3 presents the spin-flip rate $\Gamma^{(n)}$ with
336: $x_0=\infty$ (homogeneous coupling; broken line) and with $x_0=10$ and
337: $\alpha\tau/\hbar=0$, 10, 20, and 40 (solid lines).
338: When the dephasing effect is negligible ($\tau=0$),
339: the spin-flip rate is significantly enhanced by
340: the entanglement of nuclear spins, even with finite $x_0$. The behavior
341: of $\Gamma^{(n)}$ is characterized by the effective number of nuclear
342: spins which couple to the electron spin ($N \rightarrow N_{\rm eff}
343: \approx x_0$ in the previous discussion).
344: With an increase in the residence time of electrons in the dot, $\tau$,
345: the enhancement of
346: $\Gamma^{(n)}$ becomes less prominent, and is negligible when $\tau$ is
347: larger than the dephasing time of nuclear spins by the inhomogeneous
348: hyperfine field.
349:
350: %%%%%%%%%%%%%%
351: \begin{figure}[hbt]
352: \begin{center}
353: \includegraphics[width=7cm]{fig3.eps}
354: \end{center}
355: \caption{Spin-flip rate $\Gamma^{(n)}$ as a function of $n$ (number
356: of transported electrons accompanied by spin flip),
357: when the hyperfine coupling constants are
358: $\alpha_k \propto \exp{\left[ -(k-1) / x_0 \right]}$ ($k=1,2,\cdots,N$;
359: $\frac{1}{N}\sum_k\alpha_k=\alpha$). $N=14$ with $x_0=10$ (solid lines) or
360: $\infty$ (broken line). The residence time of the first electron
361: in the dot, $\tau$, is (a) $\alpha\tau/\hbar = 0$, (b) 10, (c) 20, and (d) 40.
362: $\Gamma^{(n)}$ is averaged over the random distribution of initial
363: states. The error bars denote the variation of $\Gamma^{(n)}$.}
364: \end{figure}
365: %%%%%%%%%%%%%%
366:
367: In actual quantum dots, % (i)
368: there are other origins of the dephasing of nuclear spins, {\it e.g}.,
369: dipole-dipole interaction. Generally, as long as the dephasing
370: time $T_2^*$ exceeds $\tau$, our mechanism
371: appears robust against the dephasing effects because the spin flip of
372: electrons drives the nuclear spins to be correlated every time.
373:
374: Our model is relevant to the experimental situation of double quantum dots
375: shown in Fig.\ 4. A spin blockade takes place when the spin of an incident
376: electron in dot $L$ is parallel to that of an electron trapped in dot
377: $R$.\cite{Ono}
378: We assume a magnetic field in which the Zeeman energy for electrons,
379: $E_{\rm Z}$, is much larger than $\alpha$ in $H_{\rm hf}$, whereas
380: the Zeeman energy for nuclear spins
381: is negligible. The electron-phonon interaction is taken into account for
382: the energy conservation in the spin flip.\cite{Sigurdur,Sigurdur2}
383: The Hamiltonian in the dots is
384: $
385: H = H_{\rm el}+H_{\rm ph}+H_{\rm T}+H_{\rm hf}+H_{\rm el-ph},
386: $
387: where
388: \begin{eqnarray*}
389: H_{\rm el} & = & \sum_{j=L,R}
390: \left[ \sum_{\sigma=\uparrow,\downarrow}(\varepsilon_{j}-E_{\rm Z}S_z)
391: n_{j,\sigma}+U n_{j,\uparrow}n_{j,\downarrow} \right], \\
392: H_{\rm ph} & = & \sum_{\mib q}\hbar\omega_{\mib q} b^{\dagger}_{\mib q}b_{\mib q},
393: \\
394: H_{\rm T} & = & T\sum_{\sigma=\uparrow,\downarrow}
395: (a^{\dagger}_{R,\sigma}a_{L,\sigma}+a^{\dagger}_{L,\sigma}a_{R,\sigma}), \\
396: H_{\rm el-ph} & = & \sum_{\mib q} (\beta_{L,\mib q}n_{L}+\beta_{R,\mib q}n_{R})
397: (b^{\dagger}_{\mib q}+b_{- \mib q}),
398: \end{eqnarray*}
399: where $a^{\dagger}_{j,\sigma}$, $a_{j,\sigma}$
400: ($b^{\dagger}_{\mib q}$, $b_{\mib q}$)
401: are creation and annihilation operators, respectively,
402: for an electron in dot $j$
403: (a phonon), $n_{j,\sigma}=a^{\dagger}_{j,\sigma} a_{j,\sigma}$
404: and $n_{j}=n_{j,\uparrow}+n_{j,\downarrow}$.\cite{Brandes0}
405:
406: In the spin blockade of $|L \uparrow \rangle |R \uparrow \rangle$,
407: an electron tunneling to $|R \downarrow \rangle |R \uparrow \rangle$
408: is accompanied by the spin flip in one of the dots, {\it e.g}.,
409: dot $L$, and emission
410: of a phonon of $\hbar\omega_{\mib q}=\delta E=
411: \varepsilon_{L}-(\varepsilon_{R}+U)-E_{\rm Z}$. The transition rate is
412: $\Gamma=A |\langle \Psi_{\rm fin}^{(L)}; \downarrow
413: |H_{\rm hf}| \Psi_{\rm init}^{(L)}; \uparrow \rangle |^2$ with
414: \[ %begin{equation}
415: A=
416: \frac{2\pi}{\hbar}
417: \left( T \frac{\beta_{L,\mib q}-\beta_{R,\mib q}}{E_{\rm Z} \delta E} \right)^2
418: D_{\rm ph}(\delta E) \left[ N_{\rm ph}(\delta E)+1 \right]
419: \] %\label{eq:A} \end{equation}
420: in the lowest order of $H_{\rm T}+H_{\rm hf}+H_{\rm el-ph}$.
421: Here, $D_{\rm ph}$ is the density of states for phonons and $N_{\rm ph}$
422: is the number of phonons.\cite{com2}
423: %This is the situation of our model.
424: This agrees with eq.\ (\ref{eq:Gamma}).
425:
426: %%%%%%%%%%%%%%
427: \begin{figure}[hbt]
428: \begin{center}
429: \includegraphics[width=6cm]{fig4.eps}
430: \end{center}
431: \caption{An experimental situation of double quantum dots.
432: A spin blockade takes place when the spin of an incident electron in
433: dot $L$ is parallel to that of the electron trapped in dot $R$.\cite{Ono}
434: The leakage current in the spin-blocked region is relevant to our theory.
435: }
436: \end{figure}
437: %%%%%%%%%%%%%%
438:
439: Besides eq.\ (\ref{eq:Gamma}), we have assumed that the nuclear spin state is
440: a pure state after an electron tunnels out from dot state
441: $|R \downarrow \rangle |R \uparrow \rangle$ to lead $R$ and is detected by
442: current measurement.\cite{com0}
443: For this assumption to be valid, the following higher-order tunneling processes
444: (cotunneling)\cite{Averin,Eto,Fujisawa} must be disregarded:
445: (i) third-order processes in which an electron tunnels out from dot $R$ to lead
446: $R$, an electron tunnels from dot $L$ to dot $R$, and finally an electron
447: tunnels into dot $L$ from lead $L$, or (ii) second-order processes in which
448: an electron tunnels out from dot $L$ ($R$) to lead $L$ ($R$), and another
449: electron tunnels in from lead $L$ ($R$) to dot $L$ ($R$).
450: In the presence of such processes and hyperfine couplings,
451: the nuclear spin state would be entangled with electron states in the leads.
452: Then our calculations would no longer be justified.
453:
454: We have also neglected the spin-orbit (SO) interaction, which
455: may play a role in the spin relaxation of electrons in quantum
456: dots.\cite{Khaetskii1,Khaetskii2} The coexistence of hyperfine and SO
457: interactions would complicate the evolution of the nuclear spin state
458: although the state remains pure (a linear combination of spin-flipped and
459: -unflipped states) in this case.
460: Its analysis is beyond the scope of the present study.
461:
462: % {\it Discussion}:
463: In conclusion,
464: we have proposed a current-induced entanglement of nuclear spins in a
465: quantum dot. The current accompanied by spin flip in the quantum dot
466: gradually increases the correlation among nuclear spins when the
467: residence time of electrons in the dot does not exceed the dephasing time
468: of nuclear spins $T_2^*$. The correlated state of nuclear spins
469: significantly enhances the spin-flip rate of electrons.
470: A relevant situation to this theory is the leakage current in
471: spin-blocked regions.
472:
473: The time dependence of the leakage current
474: $I(t)$ shown in Fig.\ 2(b) may be observed in the experiment
475: using a sequence
476: of NMR pulses. The nuclear spins are randomized by each pulse.
477: By tuning the interval of the pulses, $\Delta t$, $I(\Delta t)$ can be
478: measured.
479: %
480: Quantitatively, the dephasing time of nuclear spins is estimated to be
481: $T_2^* \sim 1$ ns in a case of GaAs quantum dots.\cite{Merkulov}
482: The saturation time of $I(t)$ is of the order of the residence time of
483: electrons in quantum dots [$t_{\rm sat}=\ln(N/2) \tau$ in eq.\
484: (\ref{eq:current})], which should be shorter than $T_2^*$.
485:
486: Finally, we comment on an analogy between our mechanism and the Dicke effect of
487: superradiance.\cite{Dicke,Brandes} In the spontaneous emission of
488: photons from $N$ atoms with two levels (pseudo-spin $S_z=\pm 1/2$),
489: all the atoms could interact with a common electromagnetic field.
490: The emission of photons is significantly enhanced if $N$ atoms are
491: excited initially. This is due to the formation of the pseudo-spin state
492: $|J,M \rangle$ with $J=N/2$. Starting from $|J,J \rangle$,
493: the state of $N$ atoms changes like a cascade, $|J,J \rangle$,
494: $|J,J-1 \rangle, \cdots$, emitting photons rapidly.
495: A similar effect has been proposed for the emission of phonons from $N$
496: equivalent quantum dots.\cite{Brandes,Brandes2}
497: The atoms (quantum dots) correspond to the nuclear spins in our model, and
498: the emission of photons (phonons) to the spin flip of electrons.
499: A major difference is the initialization.
500: $N$ excited states must be prepared by pumping in the Dicke effect,
501: whereas such initialization is not necessary in our mechanism.
502:
503: The authors are indebted to L.\ P.\ Kouwenhoven for the suggestion of
504: the experiment using NMR pulses. They gratefully acknowledge discussions with
505: K.\ Kawamura, R.\ Fukuda, S.\ Komiyama, T.\ Inoshita, A.\ Shimizu,
506: K.\ Ono, G.\ E.\ W.\ Bauer, and S.\ I.\ Erlingsson.
507: This work was partially supported by a Grant-in-Aid for
508: Scientific Research in Priority Areas ``Semiconductor Nanospintronics''
509: (No.\ 14076216) of the Ministry of Education, Culture, Sports, Science
510: and Technology, Japan.
511:
512: \begin{thebibliography}{99}
513: \bibitem{Loss}
514: D.\ Loss and D.\ P.\ DiVincenzo: Phys.\ Rev.\ A {\bf 57} (1998) 120.
515: \bibitem{opt1}
516: M.\ Paillard, X.\ Marie, P.\ Renucci, T.\ Amand, A.\ Jbeli and
517: J.\ M.\ Gerard: Phys.\ Rev.\ Lett.\ {\bf 86} (2001) 1634.
518: \bibitem{opt2}
519: A.\ S.\ Lenihan, M.\ V.\ Gurudev Dutt and D.\ G.\ Steel:
520: Phys.\ Rev.\ Lett.\ {\bf 88} (2002) 223601.
521: \bibitem{Fujisawa}
522: T.\ Fujisawa, D.\ G.\ Austing, Y.\ Tokura, Y.\ Hirayama and S.\ Tarucha:
523: Nature {\bf 419} (2002) 278.
524: \bibitem{Ono}
525: K.\ Ono, D.\ G.\ Austing, Y.\ Tokura and S.\ Tarucha:
526: Science {\bf 297} (2002) 1313.
527: \bibitem{Sigurdur}
528: S.\ I.\ Erlingsson, Yu.\ V.\ Nazarov and V.\ I.\ Fal'ko:
529: Phys.\ Rev.\ B {\bf 64} (2001) 195306.
530: \bibitem{Sigurdur2}
531: S.\ I.\ Erlingsson and Yu.\ V.\ Nazarov:
532: Phys.\ Rev.\ B {\bf 66} (2002) 155327.
533: \bibitem{Merkulov}
534: I.\ A.\ Merkulov, Al.\ L.\ Efros and M.\ Rosen:
535: Phys.\ Rev.\ B {\bf 65} (2002) 205309.
536: \bibitem{Lyanda-Geller}
537: Y.\ B.\ Lyanda-Geller, I.\ L.\ Aleiner and B.\ L.\ Altshuler:
538: Phys.\ Rev.\ Lett.\ {\bf 89} (2002) 107602.
539: \bibitem{Khaetskii}
540: A.\ V.\ Khaetskii, D.\ Loss and L.\ Glazman:
541: Phys.\ Rev.\ Lett {\bf 88} (2002) 186802.
542: \bibitem{Schliemann}
543: J.\ Schliemann, A.\ V.\ Khaetskii and D.\ Loss:
544: Phys.\ Rev.\ B {\bf 66} (2002) 245303.
545: \bibitem{Khaetskii1}
546: A.\ V.\ Khaetskii and Yu.\ V.\ Nazarov:
547: Phys.\ Rev.\ B {\bf 61} (2000) 12639.
548: \bibitem{Khaetskii2}
549: A.\ V.\ Khaetskii and Yu.\ V.\ Nazarov:
550: Phys.\ Rev.\ B {\bf 64} (2001) 125316.
551: \bibitem{Averin}
552: D.\ V.\ Averin and Yu.\ V.\ Nazarov:
553: Phys.\ Rev.\ Lett.\ {\bf 65} (1990) 2446.
554: \bibitem{Eto}
555: M.\ Eto: Jpn.\ J.\ Appl.\ Phys.\ {\bf 40} (2001) 1929.
556: %-------------
557: \bibitem{DNP1}
558: B.\ E.\ Kane, L.\ N.\ Pfeiffer and K.\ W.\ West:
559: Phys.\ Rev.\ B {\bf 46} (1992) 7264.
560: \bibitem{DNP2}
561: K.\ R.\ Wald, L.\ P.\ Kouwenhoven, P.\ L.\ McEuen, N.\ C.\ van der Vaart
562: and C.\ T.\ Foxon: Phys.\ Rev.\ Lett.\ {\bf 73} (1994) 1011.
563: \bibitem{DNP3}
564: D.\ C.\ Dixon, K.\ R.\ Wald, P.\ L.\ McEuen and M.\ R.\ Melloch:
565: Phys.\ Rev.\ B {\bf 56} (1997) 4743.
566: \bibitem{Machida}
567: T.\ Machida, S.\ Ishizuka, T.\ Yamazaki, S.\ Komiyama, K.\ Muraki and
568: Y.\ Hirayama: Phys.\ Rev.\ B {\bf 65} (2002) 233304.
569: \bibitem{Kronmuller}
570: S.\ Kronm\"uller, W.\ Dietsche, K.\ von Klitzing, G.\ Denninger,
571: W.\ Wegscheider and M.\ Bichler: Phys.\ Rev.\ Lett.\ {\bf 82} (1999) 4070.
572: \bibitem{Smet}
573: J.\ H.\ Smet, R.\ A.\ Deutschmann, W.\ Wegscheider, G.\ Abstreiter and
574: K.\ von Klitzing: Phys.\ Rev.\ Lett.\ {\bf 86} (2001) 2412.
575: %-------------
576: \bibitem{com0}
577: We assume that the final state of nuclear spins is a pure
578: state, not a mixed state with electron states $| \uparrow \rangle$ or
579: $| \downarrow \rangle$, after the electron tunnels out of the dot.
580: We observe a spin flip of electron by the leakage current in a spin-blocked
581: situation, whereas we cannot observe which of the nuclear spins was flipped.
582: In a spin-flip process, a phonon may be emitted. However, if the wavelength
583: of the phonon is much longer than the nucleus-nucleus distance,
584: we do not obtain such information on the nuclear spins.
585: \bibitem{com1}
586: $\Psi^{(0)}$ is different from
587: $\Psi^{(0) \prime}=
588: (c_1 |1/2 \rangle_1 +d_1 |-1/2 \rangle_1)
589: \otimes \cdots \otimes (c_N |1/2 \rangle_N+d_N |-1/2 \rangle_N)$,
590: with random coefficients $\{ c_k, d_k \}$,
591: in which the nuclear spins are not entangled at all.
592: Numerical calculations using $\Psi^{(0) \prime}$, however, yield
593: qualitatively the same results as those in Fig.\ 3 obtained using $\Psi^{(0)}$.
594: See M.\ Eto, T.\ Ashiwa and M.\ Murata: in preparation.
595: \bibitem{com3}
596: Hence the entanglement of nuclear spins cannot be observed
597: by optical experiments in which the spin relaxation of single excitations
598: is measured.\cite{opt1,opt2}
599: \bibitem{Brandes0}
600: A similar model has been studied by
601: T.\ Brandes and B.\ Kramer: Phys.\ Rev.\ Lett.\ {\bf 83} (1999) 3021.
602: \bibitem{com2}
603: For the transition rate from $|L \downarrow \rangle |R \downarrow \rangle$
604: to $|R \uparrow \rangle |R \downarrow \rangle$, $\delta E$ is replaced
605: by $\delta E'=\delta E+2E_{\rm Z}$. Note that
606: the spin of an incident electron in dot $L$ is parallel or anti-parallel
607: to the spin of an electron trapped in dot $R$, with equal probability.
608: In the latter case,
609: the spin blockade does not occur; the electron tunnels in the second dot
610: and finally out to lead $R$ immediately. The current is the sum of the
611: currents with and without spin flip.
612: \bibitem{Dicke}
613: R.\ H.\ Dicke: Phys.\ Rev.\ {\bf 93} (1954) 99.
614: \bibitem{Brandes}
615: T.\ Brandes: {\it Interacting Electrons in Nanostructures}, eds.\
616: R.\ Haug and H.\ Schoeller (Springer, Berlin Heidelberg, 2001).
617: \bibitem{Brandes2}
618: T.\ Brandes, J.\ Inoue and A.\ Shimizu:
619: Phys.\ Rev.\ Lett.\ {\bf 80} (1998) 3952.
620: \end{thebibliography}
621:
622: \end{document}
623: