1: %\documentstyle[preprint,aps, 12pt]{revtex}
2: %\documentclass[prb,showpacs,12pt]{revtex4}
3:
4:
5: \documentclass[twocolumn,prb,showpacs]{revtex4}
6: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
7: \usepackage{amsmath}
8: \usepackage{amssymb}
9: \usepackage{graphicx}
10:
11: \setcounter{MaxMatrixCols}{10}
12: %TCIDATA{OutputFilter=Latex.dll}
13: %TCIDATA{Version=4.00.0.2321}
14: %TCIDATA{Created=Thu Jun 26 16:42:24 2003}
15: %TCIDATA{LastRevised=Sunday, October 17, 2004 16:22:39}
16: %TCIDATA{<META NAME="GraphicsSave" CONTENT="32">}
17: %TCIDATA{Language=American English}
18:
19: \newtheorem{theorem}{Theorem}
20: \newtheorem{acknowledgement}[theorem]{Acknowledgement}
21: \newtheorem{algorithm}[theorem]{Algorithm}
22: \newtheorem{axiom}[theorem]{Axiom}
23: \newtheorem{claim}[theorem]{Claim}
24: \newtheorem{conclusion}[theorem]{Conclusion}
25: \newtheorem{condition}[theorem]{Condition}
26: \newtheorem{conjecture}[theorem]{Conjecture}
27: \newtheorem{corollary}[theorem]{Corollary}
28: \newtheorem{criterion}[theorem]{Criterion}
29: \newtheorem{definition}[theorem]{Definition}
30: \newtheorem{example}[theorem]{Example}
31: \newtheorem{exercise}[theorem]{Exercise}
32: \newtheorem{lemma}[theorem]{Lemma}
33: \newtheorem{notation}[theorem]{Notation}
34: \newtheorem{problem}[theorem]{Problem}
35: \newtheorem{proposition}[theorem]{Proposition}
36: \newtheorem{remark}[theorem]{Remark}
37: \newtheorem{solution}[theorem]{Solution}
38: \newtheorem{summary}[theorem]{Summary}
39: \input{tcilatex}
40:
41: \begin{document}
42:
43: \title{Effects of spin imbalance on the electric-field driven quantum
44: dissipationless spin current in $p$-doped Semiconductors}
45: \author{Liangbin Hu$^{1,2}$, Ju Gao$^1$, and Shun-Qing Shen$^1$ }
46: \affiliation{$^1$Department of Physics, The University of Hong Kong, Pokfulam Road, Hong
47: Kong, P.R.China \\
48: $^2$Department of Physics, South China Normal University, Guangdong 510631,
49: P.R.China}
50: \date{October 18, 2004}
51: \begin{abstract}
52: It was proposed recently by Murakami et al. [Science \textbf{301},
53: 1348(2003)] that in a large class of $p$-doped semiconductors, an applied
54: electric field can drive a quantum dissipationless spin current in the
55: direction perpendicular to the electric field. In this paper we investigate
56: the effects of spin imbalance on this intrinsic $spin$ Hall effect. We show
57: that in a real sample with boundaries, due to the presence of spin imbalance
58: near the edges of the sample, the spin Hall conductivity is not a constant
59: but a sensitively $position$-$dependent$ quantity, and due to this fact, in
60: order to take the effects of spin imbalance properly into account, a
61: microscopic calculation of both the quantum dissipationless spin\ Hall
62: current and the spin accumulation on an equal footing is thus required.
63: Based on such a microscopic calculation, a detailed discussion of the
64: effects of spin imbalance on the intrinsic spin Hall effect in thin slabs of
65: $p$-doped semiconductors are presented.
66: \end{abstract}
67:
68: \pacs{73.43.-f, 72.25.Dc, 72.25.Hg, 85.75.-d}
69: \maketitle
70:
71: %\date{\today}
72:
73: \section{Introduction}
74:
75: Efficient injection and coherent control of spins in non-magnetic
76: semiconductors represent two principal challenges in the emerging field of
77: spintronics, a new paradigm of semiconductor electronics based on the
78: utilization of the electron's spin degree of freedom\cite{Prinz98}. At a
79: first glance, it seems a trivial thing to inject spins into non-magnetic
80: semiconductors by use of ferromagnetic metals as sources. However, in
81: reality it is not practical because most of the spin polarizations will be
82: lost at the interface between metal and semiconductor due to the large
83: conductivity mismatch\cite{Hammar99,Schmidt00}. A possible approach that can
84: solve this problem is to replace ferromagnetic metals by ferromagnetic
85: semiconductors ( such as Ga$_{1-x}$Mn$_{x}$As ) as sources of spin injection%
86: \cite{Ohno99,Fiederling99,Mattana03}, but for practical use at room
87: temperature, the Curie temperatures of ferromagnetic semiconductors are
88: still too low. Due to such difficulties, how to achieve efficient injection
89: of spins into non-magnetic semiconductors at room temperature remains an
90: open question and more great efforts are still needed. Recently, based on
91: the Luttinger effective Hamiltonian\cite{Luttinger}, Murakami et al.
92: theoretically predicted that an extraordinary $spin$ Hall effect may occur
93: in a large class of $p$-doped semiconductors ( such as Si, Ge, and GaAs ),
94: which means that in such a semiconductor, an applied electric field can
95: drive a substantial amount of quantum dissipationless spin current in the
96: direction perpendicular to the electric field, and the spin current does not
97: decrease substantially even at room temperature\cite{Murakami03, Murakami2}.
98: This effect might reveal a new way for achieving efficient spin injection in
99: non-magnetic semiconductors at room temperature without the need of
100: ferromagnetic metals and may also find some other important applications in
101: spintronics. Prior to the discovery of this effect, a similar effect was
102: also predicted by Hirsch\cite{Hirsch99} and discussed extensively by several
103: other authors\cite{Zhang00, Hu03}. From the theoretical points of view, the
104: effect conceived by Hirsch is an $extrinsic$ spin Hall effect, which is
105: caused by the spin-orbit dependent anisotropic scatterings from impurities
106: but not an intrinsic property of a material, and it will disappear
107: completely in the absence of impurity scatterings. The spin current
108: generated by the extrinsic spin Hall effect was shown to be rather small\cite%
109: {Hirsch99, Zhang00, Hu03}, so it is of little use in the problem of spin
110: injections in non-magnetic semiconductors. Unlike the extrinsic spin Hall
111: effect, the spin Hall effect proposed in Refs.[8-9] is purely $intrinsic$,
112: which arises from the intrinsic spin-orbit coupling in the valence bands of $%
113: p$-doped semiconductors and does not rely on any spin-orbit dependent
114: anisotropic scatterings from impurities. From a more profound point of view,
115: this effect has a deep topological character and shares some basic features
116: with the quantum Hall edge current both physically and mathematically\cite%
117: {Murakami03, Murakami2}. For example, just like the case of quantum Hall
118: effect\cite{Thouless, Hatsugai, Niu}, the spin Hall conductivity due to this
119: effect is a dissipationless transport coefficient and can be expressed as an
120: integral over all states below the Fermi energy, and the contribution of
121: each state can be expressed entirely in terms of the curvature of a gauge
122: field in momentum space\cite{Murakami03, Murakami2}. Due to such features,
123: the spin current generated by this intrinsic spin Hall effect can be very
124: large ( comparable to the ordinary charge currents) and hence can serve as
125: an effective source for efficient injections of spins in non-magnetic
126: semiconductors at room temperature. Very recently, a similar intrinsic spin
127: Hall effect was also found by Sinova et al. in two-dimensional electron
128: gases ( 2DEGs) with Rashba spin-orbit coupling\cite{Sinova}. They found that
129: \ in 2DEGs with Rashba spin-orbit coupling, the dissipationless and
130: intrinsic spin Hall conductivity will take a universal value as long as both
131: spin-orbit split bands are occupied. It is anticipated this effect will also
132: find some important applications in the emerging field of spintronics.
133:
134: Although some basic concepts about the intrinsic spin Hall effect are clear%
135: \cite{Murakami03, Murakami2, Sinova}, there are still a number of important
136: questions which are needed to be further clarified, and in the last year
137: many theoretical works have been devoted to the study of this extraordinary
138: effect.\cite{Shen, Schliemann, Rashba, Murakami04, Culcer, Sinitsyn, Jphu,
139: Inoue, Dimitrova, Burkov} Basically, most of these theoretical works have
140: been focused on the calculation of the intrinsic spin Hall conductivity. In
141: the present paper, we present a theoretical investigation on the effects of
142: spin imbalance on the intrinsic spin Hall effect in $p$-doped
143: semiconductors. While it was well known both experimentally and
144: theoretically that in spin-polarized transport phenomena\cite%
145: {Pratt91,Gijs93, Johnson85, Valet93, Heide} ( including in semiconductor
146: spintronics devices\cite{Awschalom, Kato,Stephens} ) spin imbalance may have
147: significant influences on the transports of spins, what influences spin
148: imbalance will have on the intrinsic spin Hall effect is still a new subject
149: and has not yet been explored. For the intrinsic spin Hall effect, from both
150: the experimental and theoretical points of view, a clear understanding of
151: the effects of spin imbalance would be much desirable because spin imbalance
152: may not only have some significant influences on the electric-field driven
153: quantum dissipationless spin current and on its practical applications but
154: also play a crucial role in the experimental measurement of the effect\cite%
155: {Murakami03}. In this paper, based on a solid microscopic ground, we will
156: derive a set of self-consistent spin transport equations which will present
157: a proper description on the interplay between the spin imbalance and the
158: electric-field driven quantum dissipationless spin current in the intrinsic
159: spin Hall effect in $p$-doped semiconductors. Starting from these spin
160: transport equations and with the help of appropriate boundary conditions,
161: the quantum dissipationless spin current and the induced nonequilibrium spin
162: accumulation in an actual sample with boundaries can be calculated
163: simultaneously on an equal footing. Our results show that the
164: characteristics of the interplay between the quantum dissipationless spin
165: current and the spin imbalance in the intrinsic spin Hall effect in $p$%
166: -doped semiconductors are very different from what was found in usual
167: spin-polarized transport phenomena ( including in the extrinsic spin Hall
168: effect ), and some usual concepts about the interplay between spin current
169: and spin imbalance cannot be applied to the intrinsic spin Hall effect.
170:
171: The paper is organized as follows: In Sec.II, we will present a microscopic
172: derivation of the spin transport equations for describing the intrinsic spin
173: Hall effect in $p$-doped semiconductors. In our derivation, the effects of
174: spin imbalance will be included explicitly. In Sec.III, by solving these
175: spin transport equations with the help of appropriate boundary conditions,
176: the electric-field driven quantum dissipationless spin current and the
177: induced nonequilibrium spin accumulation in thin slabs of $p$-doped
178: semiconductors will be calculated explicitly.
179:
180: \section{Spin transport equations in the presence of spin imbalance}
181:
182: In a large class of $p$-doped semiconductors such as Si, Ge, and GaAs, the
183: valence bands are fourfold degenerate at the $\Gamma $ point. In the
184: momentum representation and taking the hole picture, the valance bands in
185: such semiconductors can be described by the following Luttinger effective
186: Hamiltonian\cite{Luttinger, Murakami03}
187:
188: \begin{equation}
189: \hat{H}_{0}=\frac{\hbar ^{2}}{2m}[(\gamma _{1}+\frac{5}{2}\gamma _{2})%
190: \mathbf{k}^{2}-2\gamma _{2}(\mathbf{k}\cdot \mathbf{S})^{2}],
191: \end{equation}%
192: where $\mathit{S}_{i}$ is the spin-3/2 matrix, $\gamma _{1}$ and $\gamma _{2%
193: \text{ }}$are the Luttinger parameters. For a given wave vector $\mathbf{k}$
194: , the Hamiltonian (1) has two eigenvalues, given by
195: \begin{eqnarray}
196: \epsilon _{H}(\mathbf{k}) &=&\epsilon _{\lambda =\pm 3/2}(\mathbf{k})=\frac{%
197: \gamma _{1}-2\gamma _{2}}{2m}\hbar ^{2}k^{2}\equiv \frac{\hbar ^{2}k^{2}}{%
198: 2m_{H}}, \\
199: \epsilon _{L}(\mathbf{k}) &=&\epsilon _{\lambda =\pm 1/2}(\mathbf{k})=\frac{%
200: \gamma _{1}+2\gamma _{2}}{2m}\hbar ^{2}k^{2}\equiv \frac{\hbar ^{2}k^{2}}{%
201: 2m_{L}},
202: \end{eqnarray}%
203: where $\lambda \equiv \hbar ^{-1}\mathbf{k}\cdot \mathbf{S}/k$ is a good
204: quantum number of the Hamiltonian $\hat{H}_{0}$. The hole bands described by
205: Eqs.(2-3) are referred to as the light-hole (LH) and heavy-hole (HH) bands,
206: respectively. When a uniform electric field $\mathbf{E}$ is applied, the
207: full Hamiltonian will be given by $\hat{H}=\hat{H}_{0}+e\mathbf{E}\cdot
208: \mathbf{x}$, where $-e$ is the charge of an electron. The equation of motion
209: for the light and heavy holes in a uniform electric field has been derived
210: in much detail in Ref.[8], and in the semiclassical approximation ( i.e.,
211: the spin is treated as a classical variable and hence commutes with the
212: current operator ), the following equation of motion was obtained therein,
213: \begin{equation}
214: \dot{k}_{i}=\frac{eE_{i}}{\hbar },\;\dot{x}_{i}=\frac{\hbar k_{i}}{%
215: m_{\lambda }}+\epsilon _{ijl}\lambda (2\lambda ^{2}-\frac{7}{2})\frac{k_{l}}{%
216: k^{3}}\dot{k}_{j},
217: \end{equation}%
218: where $\epsilon _{ijl}$ is the usual fully antisymmetric tensor in three
219: dimensions. The occurrence of the last term in Eq.(4) is unusual, it
220: represents a \textquotedblleft Lorentz force\textquotedblright\ in momentum
221: space and is a natural generalization of the quantum Hall effect\cite%
222: {Thouless, Hatsugai, Niu} to three dimensions\cite{Murakami03, Murakami2}.
223: It is just due to this \textquotedblleft Lorentz force\textquotedblright\ in
224: momentum space ( which makes the hole velocity noncollinear with its
225: momentum ) that the applied electric field will drive a quantum
226: dissipationless spin Hall current in the direction perpendicular to the
227: electric field. From Eq.(4), one can get that in the low temperature limit
228: and in the semiclassical approximation, the net spin current due to both the
229: LH and HH bands will be given by\cite{Murakami03}
230: \begin{eqnarray}
231: J_{j}^{_{i}} &=&\frac{\hbar }{3}\sum_{\lambda ,\mathbf{k}}\dot{x}_{j}\frac{%
232: \lambda k_{i}}{k}n_{\lambda }(\mathbf{k}) \notag \\
233: &=&\sigma _{s}^{0}\epsilon _{ijk}E_{k},
234: \end{eqnarray}%
235: where $J_{j}^{_{i}}$ denotes the net spin current following to the $x_{j}$
236: direction with spin parallel to the $x_{i}$ direction, $n_{\lambda }(\mathbf{%
237: k})$ is the filling of holes in the band with helicity $\lambda $, and $%
238: \sigma _{s}^{0}$ is the $spin$ Hall conductivity, which is given by
239: \begin{equation}
240: \sigma _{s}^{0}=\frac{e}{12\pi ^{2}}(3k_{H}^{F}-k_{L}^{F}),
241: \end{equation}%
242: with $k_{L}^{F}$ and $k_{H}^{F}$ denoting the Fermi wave numbers in the LH
243: and HH bands, respectively. In obtaining Eqs.(5-6), one has assumed that the
244: fillings of holes in each band can be described by the simple Fermi-Dirac
245: equilibrium distribution function. An alternative way of calculating the
246: intrinsic spin Hall conductivity is by use of the Kubo formula\cite%
247: {Murakami2, Sinova, Shen, Schliemann, Culcer, Sinitsyn}. Based on the Kubo
248: formula, as was shown in Ref.[9], the full quantum treatment of the
249: noncommutativity between the quantum spin and current operator will lead to
250: a quantum correction to Eq.(6). But if one takes the semiclassical limit,
251: the result will become the same as was given by Eq.(6).
252:
253: Eqs.(5-6) are the central results of Refs.[8-9]. They are valid in the
254: absence of spin imbalance. But in a real sample with boundaries, when a spin
255: current circulates in it, spin imbalance will be caused inevitably near the
256: edge of the sample by the spin current, and in the presence of spin
257: imbalance, the spin current may be significantly different from what was
258: given by Eqs.(5-6), especially in the regions near the edges of the sample.
259: The reason for this is that in the presence of spin imbalance, the fillings
260: of holes in the LH and HH bands may deviate significantly from the
261: corresponding cases in the equilibrium state. In order to take the effects
262: of spin imbalance properly into account, one should obtain the distribution
263: function strictly by solving the Boltzman transport equation, which
264: describes the changes of the distribution function in a nonequilibrium
265: state. In a nonequilibrium but steady state, the Boltzman equation reads
266: \begin{equation}
267: \mathbf{\dot{x}}\cdot \nabla f_{\lambda }(\mathbf{x},\mathbf{k})+\mathbf{%
268: \dot{k}}\cdot \nabla _{\mathbf{k}}f_{\lambda }(\mathbf{x},\mathbf{k}%
269: )=-\sum_{\lambda ^{^{\prime }}}(\frac{\partial f_{\lambda }}{\partial t}%
270: )_{\lambda \rightarrow \lambda ^{^{\prime }}}^{(coll.)},
271: \end{equation}%
272: where $(\partial f_{\lambda }/\partial t)_{\lambda \rightarrow \lambda
273: ^{^{\prime }}}^{(coll.)}$ is the collision term due to impurity scatterings,
274: and $\mathbf{\dot{x}}$ and $\mathbf{\dot{k}}$ are the drift velocities of
275: holes in the real space and in the momentum space, respectively. Similar to
276: Ref.[8], in this paper we will confine our discussion to the semiclassical
277: limit and weak external electric field ( i.e., in the linear response regime
278: ) so that the semiclassical equation of motion given by Eq.(4) can be applied%
279: \cite{Murakami03, Murakami2}. The collision term $(\partial f_{\lambda
280: }/\partial t)_{\lambda \rightarrow \lambda ^{^{\prime }}}^{(coll.)}$ will be
281: given by
282: \begin{eqnarray}
283: (\frac{\partial f_{\lambda }}{\partial t})_{\lambda \rightarrow \lambda
284: ^{^{\prime }}}^{(coll.)} & =& -\int \frac{d^{3}\mathbf{k}^{^{\prime }}}{%
285: (2\pi )^{3}}w_{\lambda ,\lambda ^{^{\prime }}}^{(i)}(\mathbf{k},\mathbf{k}%
286: ^{^{\prime }})\delta (\epsilon _{\lambda }(\mathbf{k})-\epsilon _{\lambda
287: ^{^{\prime }}}(\mathbf{k}^{^{\prime }})) \notag \\
288: & \times & \lbrack f_{\lambda }(\mathbf{x},\mathbf{k})-f_{\lambda ^{^{\prime
289: }}}(\mathbf{x},\mathbf{k}^{^{\prime }})],
290: \end{eqnarray}%
291: where $w_{\lambda ,\lambda ^{^{\prime }}}^{(i)}(\mathbf{k},\mathbf{k}%
292: ^{^{\prime }})$ is the probability of a hole to be scattered from the state $%
293: |\mathbf{k}\lambda \rangle $ into the state $|\mathbf{k}^{^{\prime }}\lambda
294: ^{^{\prime }}\rangle $ due to impurity scatterings, and the impurity
295: scatterings will be assumed to be isotropic and spin-independent.
296:
297: In the equilibrium state, the fillings of holes in each band are stable and
298: can be described by the simple Fermi-Dirac equilibrium distribution
299: function. When the external electric field is applied and the system turns
300: into an nonequilibrium but steady state, the fillings of holes in each band
301: will still be stable but different from what was described by the simple
302: Fermi-Dirac equilibrium distribution function. In the presence of spin
303: imbalance, the changes of the fillings of holes in the LH and HH bands will
304: be caused primarily by two kinds of contributions. The first kind of
305: contribution is caused by the drifts of holes in the external electric
306: field, and the second kind of contribution is due to the occurrence of spin
307: imbalance. In the linear response regime ( i.e., in a weak electric field ),
308: the deviations of the fillings of holes in each band from the corresponding
309: cases in the equilibrium state are small, and the two kinds of contributions
310: will be independent and both be \ proportional to $\partial f^{0}/\partial
311: \epsilon _{F}$ ( here $f^{0}=\{\exp [\beta (\epsilon _{\lambda }(\mathbf{k}%
312: )-\epsilon _{F})]+1\}^{-1}$ is the usual Fermi-Dirac equilibrium
313: distribution function with $\beta $ denoting the inverse of temperature and $%
314: \epsilon _{F}$ the Fermi level in the equilibrium state ). Considering this
315: fact and by use of the relaxation time approximation, in the linear response
316: regime the nonequilibrium distribution function $f_{\lambda }(\mathbf{x},%
317: \mathbf{k})$ ( in a nonequilibrium but steady state ) can be expressed as
318: the following,
319: \begin{equation}
320: f_{\lambda }(\mathbf{x},\mathbf{k})=f^{0}+\mu _{\lambda }(\mathbf{x})\frac{%
321: \partial f^{0}}{\partial \epsilon _{F}}+e\tau \lbrack \mathbf{E}_{\lambda }(%
322: \mathbf{x})\cdot \mathbf{V}_{\lambda }]\frac{\partial f^{0}}{\partial
323: \epsilon _{F}},
324: \end{equation}%
325: where $\mathbf{V}_{\lambda }=\hbar \mathbf{k/}m_{\lambda }$ is the velocity
326: of holes; $\tau $ is the total relaxation time of holes due to
327: impurity-induced random scatterings; $\mathbf{E}_{\lambda }(\mathbf{x})$ is
328: the total effective field felt by a moving hole in the band with helicity $%
329: \lambda $, which is the sum of the external electric field $\mathbf{E}$ and
330: a band-dependent effective field induced by the spin imbalance in the
331: sample. The detailed definition of $\mathbf{E}_{\lambda }(\mathbf{x})$ and $%
332: \tau $ will be given below. The second term in Eq.(9) just characterizes the
333: deviation of the filling of holes in the band with helicity $\lambda $ from
334: the corresponding case in the equilibrium state due to the occurrence of
335: spin imbalance in the sample, and the presence of this term is
336: mathematically equivalent to introducing a band-dependent \textquotedblleft\
337: shift \textquotedblright\ $\mu _{\lambda }$ in the Fermi level $\epsilon
338: _{F} $. ( It should be noted that unlike the corresponding cases in usual
339: spin-polarized transport phenomena, here $\mu _{\lambda }$ does not relate
340: directly to the spin accumulation because the label $\lambda $ does not
341: correspond to a fixed spin-polarization direction in real space. ). The
342: third term in Eq.(9) denotes the change of the filling due to the drifts of
343: holes in the external electric field and in the presence of impurity
344: scatterings. By inserting Eq.(9) into Eq.(7) and assuming that the impurity
345: scatterings are isotropic, the Boltzman equation can be simplified to the
346: following form,
347: \begin{eqnarray}
348: &&\mathbf{V}_{\lambda }\cdot \lbrack \mathbf{E}+\frac{1}{e}\bigtriangledown
349: \mu _{\lambda }(\mathbf{x})+\tau \bigtriangledown (\mathbf{E}_{\lambda }(%
350: \mathbf{x})\cdot \mathbf{V}_{\lambda })] \notag \\
351: &=&\sum_{\lambda ^{^{\prime }}(\neq \lambda )}\frac{\mu _{\lambda }(\mathbf{x%
352: })-\mu _{\lambda ^{^{\prime }}}(\mathbf{x})}{e\tau _{\lambda \lambda
353: ^{^{\prime }}}}+\sum_{\lambda ^{^{\prime }}}\frac{\tau \mathbf{E}_{\lambda }(%
354: \mathbf{x})\cdot \mathbf{V}_{\lambda }}{\tau _{\lambda \lambda ^{^{\prime }}}%
355: },
356: \end{eqnarray}%
357: where $\tau _{\lambda \lambda ^{^{\prime }}}$ is a characteristic relaxation
358: time defined by
359: \begin{equation}
360: \tau _{\lambda \lambda ^{^{\prime }}}=[\int \frac{d^{3}\mathbf{k}^{\prime }}{%
361: (2\pi )^{3}}w_{\lambda ,\lambda ^{^{\prime }}}^{(i)}(\mathbf{k},\mathbf{k}%
362: ^{^{\prime }})\delta (\epsilon _{\lambda }(\mathbf{k})-\epsilon _{\lambda
363: ^{^{\prime }}}(\mathbf{k}^{^{\prime }}))]^{-1},
364: \end{equation}%
365: which characterizes the probability ( given by $\tau _{\lambda \lambda
366: ^{^{\prime }}}^{-1}$ ) for a hole in the band with helicity $\lambda $ to be
367: scattered into the band with helicity $\lambda ^{^{\prime }}$ due to
368: impurity scatterings. For simplicity, in the following we will assume that
369: the intra-band-scattering relaxation time $\tau _{\lambda \lambda }\equiv
370: \tau _{1}$ ( independent of $\lambda $ ) and the inter-band-scattering
371: relaxation time $\tau _{\lambda \lambda ^{^{\prime }}(\neq \lambda )}\equiv
372: \tau _{2}$ ( independent of $\lambda $ and $\lambda ^{^{\prime }}$ ), and as
373: usual, the total scattering probability for a hole ( given by $\tau ^{-1}$,
374: i.e., the inverse of the total relaxation time of a hole ) can be given by
375: \begin{equation}
376: \tau ^{-1}=\sum_{\lambda ^{^{\prime }}}\tau _{\lambda \lambda ^{^{\prime
377: }}}^{-1}=\frac{1}{\tau _{1}}+\frac{3}{\tau _{2}},
378: \end{equation}%
379: which is assumed to be independent of the band label $\lambda $. Multiplying
380: both sides of Eq.(10) by $\mathbf{V}_{\lambda }$ and then integrating both
381: sides with respect to $\mathbf{V}_{\lambda }$ and with the help of Eq.(12),
382: one can find that the total effective field felt by a moving hole in the
383: band with helicity $\lambda $ should be given by
384: \begin{equation}
385: \mathbf{E}_{\lambda }(\mathbf{x})=\mathbf{E}+\frac{1}{e}\bigtriangledown \mu
386: _{\lambda }(\mathbf{x}).
387: \end{equation}%
388: Eq.(13) suggests that in the presence of spin imbalance, in addition to the
389: external electric field $\mathbf{E}$, conduction electrons will also feel an
390: effective field proportional to the gradient of the band- and
391: position-dependent shift in the Fermi level. After $\tau $ and $\mathbf{E}%
392: _{\lambda }(\mathbf{x})$ are determined from Eqs.(12-13), the nonequilibrium
393: distribution function $f_{\lambda }(\mathbf{x},\mathbf{k})$ will also be
394: determined by Eq.(9). Then in the semiclassical limit the electric-field
395: driven quantum dissipationless spin current can be obtained through the
396: following formula,
397: \begin{equation}
398: J_{j}^{i}(\mathbf{x})=\sum_{\lambda }\int \frac{d^{3}\mathbf{k}}{(2\pi )^{3}}%
399: [\dot{x}_{j}s_{i,\lambda }(\mathbf{k})]f_{\lambda }(\mathbf{x,k}),
400: \end{equation}%
401: where $\dot{x}_{j}=\hbar k_{j}/m_{\lambda }+\epsilon _{jkl}\lambda (2\lambda
402: ^{2}-7/2)k_{l}\dot{k}_{k}/k^{3}$ ( see Eq.(4) ) and $s_{i,\lambda }(\mathbf{k%
403: })=\hbar (\lambda k_{i}/k)/3$ are the velocity and the spin of a hole with
404: momentum $\mathbf{k}$ and helicity $\lambda $, respectively. ( Since the
405: spin-$3/2$ matrix $\mathbf{S}$ in the Hamiltonian (1) is a summation of the
406: spin angular momentum $\mathbf{s}$ with spin one-half and the atomic orbital
407: angular momentum $\mathbf{l}$ with spin one, the expectation value of $%
408: \mathbf{s}$ should be one third of $\mathbf{S}$.\cite{Murakami03,Luttinger}%
409: ). By substituting Eqs.(12-13) into Eq.(9) and then inserting Eq.(9) into
410: Eq.(14), the following result can be obtained
411: \begin{equation}
412: J_{j}^{i}(\mathbf{x})=\sigma _{s}(\mathbf{x})\epsilon _{ijk}E_{k},
413: \end{equation}%
414: where $\sigma _{s}(\mathbf{x})$ is the spin Hall conductivity in the
415: presence of spin imbalance, which is given by
416: \begin{equation}
417: \sigma _{s}(\mathbf{x})=\sigma _{s}^{0}-\frac{e}{48\pi ^{2}\epsilon _{F}}%
418: [3k_{H}^{F}\mu _{H}(\mathbf{x})-k_{L}^{F}\mu _{L}(\mathbf{x})].
419: \end{equation}%
420: Here $\sigma _{s}^{0}$ is the spin Hall conductivity in the $absence$ of
421: spin imbalance, which has been defined in Eq.(6), and $\mu _{H}(\mathbf{x}%
422: )\equiv \mu _{3/2}(\mathbf{x})+\mu _{-3/2}(\mathbf{x})$ and $\mu _{L}(%
423: \mathbf{x})\equiv \mu _{1/2}(\mathbf{x})+\mu _{-1/2}(\mathbf{x})$.
424: Eqs.(15-16) show that the effects of spin imbalance on the quantum
425: dissipationless spin current due to the intrinsic spin Hall effect in $p$%
426: -doped semiconductors are very different from what was found in usual
427: spin-polarized phenomena ( including the extrinsic spin Hall effect\cite%
428: {Hirsch99, Zhang00, Hu03} ). First, in the presence of spin imbalance, the
429: spin Hall conductivity due to the intrinsic spin Hall effect in $p$-doped
430: semiconductors might not be a constant but a $position$-$dependent$
431: quantity. This is a new feature that was not seen before. Second, for the
432: intrinsic spin Hall effect in $p$-doped semiconductors, the change of the
433: quantum dissipationless spin current due to the occurrence of spin imbalance
434: is determined directly by $\mu _{\lambda }(\mathbf{x})$ ( i.e., the
435: band-dependent \textquotedblleft\ shifts \textquotedblright\ in the Fermi
436: level ) but is independent of the gradients of $\mu _{\lambda }(\mathbf{x})$%
437: . This is also significantly different from what was found in usual
438: spin-polarized transports ( including the extrinsic spin Hall effect). These
439: unusual characteristics of the intrinsic spin Hall effect in $p$-doped
440: semiconductors can be understood by the following arguments. According to
441: Eq.(6), the spin Hall conductivity should be determined uniquely by the
442: Fermi wave numbers $k_{H}^{F}$ and $k_{L}^{F}$. In the presence of spin
443: imbalance, because the spin imbalance will induce a position and band
444: dependent shift $\mu _{\lambda }(\mathbf{x})$ in the Fermi level, the Fermi
445: wave numbers $k_{H}^{F}$ and $k_{L}^{F}$ will also be position-dependent,
446: and the changes of $k_{H}^{F}$ and $k_{L}^{F}$ due to the occurrence of spin
447: imbalance will be determined directly by $\mu _{\lambda }(\mathbf{x})$. Due
448: to this reason, in the presence of spin imbalance, the spin Hall
449: conductivity will be a position-dependent quantity, and the change of the
450: spin Hall current due to the occurrence of spin imbalance will be determined
451: by $\mu _{\lambda }(\mathbf{x})$ but independent of the gradients of $\mu
452: _{\lambda }(\mathbf{x})$. Finally, it should be pointed out that because we
453: have considered only isotropic and spinless impurity scattering, the
454: mechanism of the generation of the spin Hall current described by
455: Eqs.(15-16) is still purely intrinsic, though there are some significant
456: differences between Eq.(16) and Eq.(6). In fact, one can check that in the
457: linear response regime the impurity scattering term ( i.e., the third term )
458: in Eq.(9) does not contribute to the spin Hall conductivity given by
459: Eq.(16). This point will be more clearly seen from the results presented in
460: Sec.III. Of course, if the impurity scatterings are spin-orbit dependent,
461: then the total spin current will contain not only the intrinsic part but
462: also contain an extrinsic part due to the spin-orbit dependent impurity
463: scatterings through the mechanism proposed by Hirsch\cite{Hirsch99, Zhang00,
464: Hu03}.
465:
466: In the ordinary charge Hall effect, the charge Hall current causes $charge$
467: imbalance in a sample and results in charge accumulation. Similarly, in the
468: spin Hall effect, the spin imbalance caused by the spin Hall current will
469: result in nonequilibrium spin accumulation in a sample. Corresponding to the
470: quantum spin Hall current given by Eqs.(15-16), the nonequilibrium spin
471: accumulation induced by the quantum spin Hall current can be obtained as the
472: following,
473: \begin{eqnarray}
474: S_{i}(\mathbf{x}) &=&\sum_{\lambda }\int \frac{d^{3}\mathbf{k}}{(2\pi )^{3}}%
475: s_{i,\lambda }(\mathbf{k})f_{\lambda }(\mathbf{x,k}) \notag \\
476: &=&\epsilon _{ijl}\frac{E_{l}\hbar ^{2}}{16e\epsilon _{F}^{2}}\frac{\partial
477: }{\partial x_{j}}[C_{L}\mu _{L}(\mathbf{x})-3C_{H}\mu _{H}(\mathbf{x})],
478: \end{eqnarray}%
479: where $C_{L}=e^{2}\tau (k_{L}^{F})^{3}/6\pi ^{2}m_{L}$ and $C_{H}=e^{2}\tau
480: (k_{H}^{F})^{3}/6\pi ^{2}m_{H}$ are the ordinary charge conductivities of
481: the light holes and the heavy holes, respectively. Eqs.(15-17) show that in
482: the intrinsic spin Hall effect in $p$-doped semiconductors, both the quantum
483: dissipationless spin current and the spin accumulation will depend
484: sensitively on $\mu _{\lambda }(\mathbf{x})$, i.e., the band-dependent
485: \textquotedblleft\ shifts \textquotedblright\ in the Fermi level. To find
486: out the equations that $\mu _{\lambda }(\mathbf{x})$ should satisfy, one can
487: substitute Eqs.(12-13) into Eq.(10) and integrate both sides of Eq.(10) with
488: respect to $\mathbf{k}$, then one will arrive at the following equation
489: \begin{equation}
490: \bigtriangledown ^{2}\mu _{\lambda }(\mathbf{x})=\frac{1}{D_{\lambda }^{2}}%
491: [4\mu _{\lambda }(\mathbf{x})-\mu _{H}(\mathbf{x})-\mu _{L}(\mathbf{x})],
492: \end{equation}%
493: where $D_{\lambda }=V_{\lambda }^{F}\sqrt{\tau _{2}\tau /3}$ is a
494: characteristic hole diffusion length and $V_{\lambda }^{F}$ is the
495: band-dependent Fermi velocity. In addition to Eq.(18), $\mu _{\lambda }(%
496: \mathbf{x})$ should also satisfy the charge neutrality condition, which
497: requires that the net changes of the charge density due to the flow of the
498: quantum spin current, given by $\delta \rho _{c}=-e\sum_{\lambda }\int \frac{%
499: d^{3}\mathbf{k}}{(2\pi )^{3}}[f_{\lambda }(\mathbf{x,k})-f^{0}(\epsilon
500: _{\lambda }(\mathbf{k}))]$, should be zero. This leads to the following
501: equation
502: \begin{equation}
503: \mu _{H}(\mathbf{x})=-(\frac{m_{L}}{m_{H}})^{3/2}\mu _{L}(\mathbf{x}).
504: \end{equation}%
505: Eqs.(15)-(19) are the central results of the present paper. They constitute
506: a set of self-consistent equations from which both the quantum
507: dissipationless spin current and the spin accumulation due to the intrinsic
508: spin Hall effect in a real sample of $p$-doped semiconductors with
509: boundaries can obtained simultaneously with the help of appropriate boundary
510: conditions.
511:
512: \section{Intrinsic spin Hall effect in thin slabs of $p$-doped semiconductors%
513: }
514:
515: Eqs.(15-19) are rather general and in principle they can be applied to
516: samples with any kind of geometries. In the experimental measurement of the
517: Hall effect ( including the spin Hall effect ), a thin slab geometry ( i.e.,
518: the $Hall$ bar ) is usually applied. In this section, starting from
519: Eqs.(15-19), we will present a detailed theoretical investigation on the
520: intrinsic spin Hall effect in a thin slab of $p$-doped semiconductors. We
521: assume that the longitudinal direction of the slab is along the $z$ axis and
522: the transverse direction along the $y$ axis and the normal of the surface
523: along the $x$ axis, respectively, and an external electric field $E_{z}$ is
524: applied in the longitudinal direction of the slab. The thickness of the slab
525: is assumed to be much smaller than the hole diffusion length $D_{\lambda }$
526: and the length of the slab is assumed to be much larger than the width, so
527: that only spin current flowing to the $y$ direction ( i.e., in the
528: transverse direction of the slab ) with spin parallel to the $x$ direction
529: need to be considered. The two boundaries of the slab are assumed to be
530: located at $y=\pm w/2$, and $w$ is the width of the slab. In general, it is
531: very difficult to solve Eqs.(15-19) analytically. In order to get some
532: explicit expressions for the spin Hall current and the spin accumulation, we
533: assume that in Eqs.(15-19) the hole diffusion length $D_{\lambda }$ is $%
534: \lambda $-independent ( i.e., $D_{\lambda }\equiv D$ ) and $m_{L}\simeq m_{H}
535: $. Then from Eqs.(18-19) one can see that $\mu _{H}(y)\equiv \mu
536: _{3/2}(y)+\mu _{-3/2}(y)$ and $\mu _{L}(y)\equiv \mu _{1/2}(y))+\mu
537: _{-1/2}(y)$ can be expressed as
538: \begin{equation}
539: \mu _{H}(y)\simeq -\mu _{L}(y)=Ae^{y/2D}+Be^{-y/2D},
540: \end{equation}%
541: where $A$\ and $B$ are two constant coefficients that need to be determined
542: by the appropriate boundary condition. In this paper, we will consider the
543: transverse open circuit boundary condition. In the transverse open circuit
544: boundary condition, the spin Hall current will be zero at the two boundaries
545: of the slab, i.e., $J_{y}^{x}(y=\pm w/2)=0$. Substituting Eq.(20) into
546: Eqs.(15-16), the spin Hall current $J_{y}^{x}(y)$ can be expressed as a
547: function of the coefficients $A$ and $B$. Then by use of the transverse open
548: circuit boundary condition, the coefficients $A$ and $B$ can be determined,
549: and one can get that
550: \begin{equation}
551: A=B=\frac{2\epsilon _{F}(3k_{H}^{F}-k_{L}^{F})}{3k_{H}^{F}+k_{L}^{F}}\frac{1%
552: }{\cosh (w/2D)}.
553: \end{equation}%
554: After the coefficients $A$ and $B$ are determined, the spin Hall current $%
555: J_{y}^{x}(y)$ and the spin Hall conductivity $\sigma _{s}(y)$ will also be
556: obtained by inserting Eq.(20) into Eqs.(15-16), and the results are given by
557: \begin{eqnarray}
558: J_{y}^{x}(y) &=&\sigma _{s}(y)E_{z}, \\
559: \sigma _{s}(y) &=&\sigma _{s}^{0}[1-\frac{\cosh (y/2D)}{\cosh (w/4D)}].
560: \end{eqnarray}%
561: Eqs.(22-23) show that, in the presence of spin imbalance, both the spin Hall
562: current and the spin Hall conductivity might be highly position-dependent
563: and might also depend sensitively on the hole diffusion length $D$ and the
564: width $w$ of the sample. The spin Hall conductivity $\sigma _{s}(y)$ will be
565: maximum at the center of the sample ( i.e., at $y=0$ ) and tend to be zero
566: at the edges of the sample. Two limiting cases will be especially
567: interesting. The first case is that the hole diffusion length $D$ is much
568: larger than the width $w$ of the sample. In this limiting case the spin Hall
569: current will be very small, i.e., $\sigma _{s}(y)\simeq 0$ everywhere. The
570: second interesting case is that $w\gg D$. In this limiting case, the maximum
571: value of the spin Hall conductivity will be given by $\sigma _{s}(y)\simeq
572: \sigma _{s}^{0}$ ( at $y=0$ ) and $\sigma _{s}(y)\rightarrow 0$ as $%
573: y\rightarrow \pm w/2$. These features can be seen clearly from Fig.1, where
574: we have plotted the position dependence of the spin Hall conductivity $%
575: \sigma _{s}(y)$ in three cases with different ratios of $w/D$. \ From Fig.1
576: and Eq.(23), one can see clearly that if no boundaries exist ( i.e., $%
577: w\rightarrow \infty $ and hence no spin imbalance occurs ), the spin Hall
578: conductivity will be a constant and return to the same result as was given
579: by Eq.(6), i.e., the spin Hall conductivity will not be changed by weak
580: isotropic and spinless impurity scatterings. This is in agreement with
581: Ref.[8] and also in agreement with the result obtained by a more accurate
582: calculation performed in Ref.[20]. It is interesting to note that recently a
583: similar conclusion was also obtained for the intrinsic spin Hall effect in
584: 2DEGs with Rashba spin-orbit coupling by both numerical simulations and
585: analytical calculations, which suggest that in the presence of weak (
586: isotropic and spinless ) impurity scatterings, the intrinsic spin Hall
587: conductivity in a Rashba two-dimensional electron gas should still take a
588: universal value, proving that the sample size exceeds the localization length%
589: \cite{Dimitrova, Burkov}. Of course, it should be pointed out that at
590: present different views also exist on this problem. For example, in Ref.[24]
591: it was argued that the spin-orbit-coupling induced intrinsic spin Hall
592: current in a Rashba two-dimensional electron gas should vanish in the
593: presence of impurity scatterings, even if the impurity scatterings are weak
594: and spinless.
595:
596: \begin{figure}[tbp]
597: \includegraphics[width=8.5cm]{figure1.eps}
598: \caption{
599: Illustration of the position dependences of the spin Hall
600: conductivity $\sigma _s(y)$ in three cases with different ratios of $w/D$. (
601: $w/D=50$ for the solid line, $w/D=10$ for the dashed line, and $w/D=1$ for
602: the dotted line. $\sigma _s(y)$ is normalized by $\sigma _s^0$, i.e., the
603: spin Hall conductivity in the absence of spin imbalance.)
604: }
605: \end{figure}
606:
607: The quantum dissipationless spin current generated by the intrinsic spin
608: Hall effect does not carry charges ( i.e., it is a $pure$ spin current), so
609: it is very difficult to measure the quantum dissipationless spin current
610: directly. An indirect but much more convenient way to detect the quantum
611: dissipationless spin current is to measure the nonequilibrium spin
612: accumulation induced by the quantum dissipationless spin current. The
613: nonequilibrium spin accumulation induced by the quantum dissipationless spin
614: Hall current in a thin slab of $p$-doped semiconductors can be got by
615: inserting Eqs.(20-21) into Eq.(17), and the following result can be
616: obtained,
617: \begin{equation}
618: S_{x}(y)=\frac{3\pi ^{2}\hbar ^{2}\sigma _{s}^{0}E_{z}(C_{L}+3C_{H})\sinh
619: (y/2D)}{e^{2}\epsilon _{F}(k_{L}^{F}+3k_{H}^{F})D\cosh (w/4D)}.
620: \end{equation}%
621: Eq.(24) shows that the spin accumulation will be linearly proportional to
622: the spin Hall conductivity $\sigma _{s}^{0}$ and also depend sensitively on
623: the ordinary charge conductivities $C_{L}$ and $C_{H}$ of the light and
624: heavy holes. It also have a sensitive dependence on the hole diffusion
625: length $D$ and the sample width $w$. \ According to Eq. (24), for a
626: infinitely large sample without boundaries ( i.e., $w\rightarrow \infty $ ),
627: no spin accumulation will appear ( i.e., $S_{x}(y)=0$ for any finite $y$ ).
628: This is different from what was found in a Rashba two-dimensional electron
629: gas, \ where it was found that the application of an in-plane electric field
630: would induce a homogeneous nonequilibrium spin accumulation without resort
631: to the boundary effects. \cite{Edelstein, Inoue2, Add} \ From Eq.(24), one
632: can see that for an actual sample with boundaries, the spatial distribution
633: of the spin accumulation due to the intrinsic spin Hall effect would be
634: highly inhomogeneous. The spin accumulation will be maximum at the edges of
635: the sample and vanish near the center of the slab, and the spin accumulation
636: at the edges of the sample will increase with the increase of the sample
637: width $w$. This has been illustrated in Fig.2. From Fig.2 one can see that
638: if the sample width $w$ is much smaller than the hole diffusion length $D$,
639: the spin accumulation induced by the quantum spin Hall current will be very
640: small. On the other hand, if the sample width $w$ is much larger than the
641: hole diffusion length $D$, the spin accumulation at the edges of the sample
642: will be almost a constant, independent of the sample width. This will be a
643: merit for the experimental measurement of the intrinsic spin Hall effect. To
644: obtain a quantitative estimation on the order of the magnitude of the spin
645: accumulation induced by the quantum dissipationless spin Hall current in a
646: real sample, let us consider some actual experimental parameters. We take
647: the ordinary conductivity $C_{L,H}\sim 10^{2}\Omega ^{-1}$cm$^{-1}$ and the
648: hole diffusion length $D\sim 10$nm and $\hbar /\epsilon _{F}\sim 1$f$\sec $.
649: These parameters are typical of the holes in GaAs with the hole density $%
650: n\sim 10^{19}$cm$^{-3}$. The width $w$ of the sample is assumed to be $%
651: 100\mu $m ( much larger than the hole diffusion length ) and a current
652: density $j_{x}\sim 10^{4}$A$/$cm$^{2}$. By use of the parameters listed
653: above, from Eq.(24) it can be estimated that the spin accumulation at the
654: edges of the sample will be on the order of $10^{13}-10^{15}\mu _{B}$cm$%
655: ^{-2} $. Such magnitudes should be large enough to be measured by some
656: ordinary experimental methods, for example, by the method proposed in
657: Refs.[8-11]. Finally, it should be pointed out that a rough estimation of
658: the spin accumulation due to the quantum dissipationless spin Hall current
659: was also presented in the supporting online material for Ref.[8] based on a
660: simple analysis by use of the usual spin diffusion equation, but there are
661: some significant differences between the results obtained in the present
662: paper and the corresponding results reported therein. This can be seen by
663: making a comparison between Eq.(24)\ obtained in the present paper and
664: Eq.(S16) presented in the supporting online material for Ref.[8]. For
665: example, according to Eq.(24) obtained in the present paper, the spin
666: accumulation will not only depend on the spin Hall conductivity but also
667: depend sensitively on the ordinary charge conductivities of the light and
668: heavy holes; however, according to Eq.(S16) in the supporting online
669: material for Ref.[8], the spin accumulation will only depend on the spin
670: Hall conductivity but is independent of the ordinary charge conductivities
671: of the light and heavy holes. Our results show that though the mechanism of
672: the intrinsic spin Hall effect is purely intrinsic, impurity scatterings
673: might have some significant influences on the effect in a real sample with
674: boundaries, and the usual spin diffusion equation might not be very suitable
675: for describing this effect. In fact, from the microscopic calculation
676: presented in Sec.II, one can see that in general the quantum dissipationless
677: spin Hall current ( given by Eqs.(15-16) ) and the spin accumulation ( given
678: by Eq.(17) ) due to the intrinsic spin Hall effect do not satisfy the usual
679: spin diffusion equation.
680:
681: \begin{figure}[tbp]
682: \includegraphics[width=8.5cm]{figure2.eps}
683: \caption{
684: Illustration of the changes of the spin accumulation at the edges of
685: a sample with the variation of the sample width. ( The spin accumulation is
686: normalized by $S_0=\frac{3\pi ^2\hbar ^2\sigma _s^0E_z(C_L+3C_H)}{%
687: e^2\epsilon _FD(k_L^F+3k_H^F)}$. )
688: }
689: \end{figure}
690:
691: In conclusion, in this paper we have presented a detailed theoretical
692: investigation on the effects of spin imbalance on the intrinsic spin Hall
693: effect in $p$-doped semiconductors. We have shown that in a real sample with
694: boundaries, the spin Hall conductivity might not be a constant but a
695: sensitively position-dependent quantity due to the occurrence of spin
696: imbalance near the edges of the sample, and in order to take the effects of
697: spin imbalance properly into account, a microscopic calculation of both the
698: quantum dissipationless spin current and the spin accumulation based on an
699: equal footing is thus required. We stress that some usual concepts about the
700: interplay between spin current and spin imbalance might not be suitable for
701: describing the intrinsic spin Hall effect. After some modifications, the
702: theory presented in this paper might also be applied to investigate the
703: effects of spin imbalance in the intrinsic spin Hall effect in 2DEGs with
704: Rashba spin-orbit coupling. Finally, it should be pointed out that though in
705: the last year many theoretical works have been devoted to the study of the
706: intrinsic spin Hall effect, many controversial issues still exist concerning
707: some fundamental aspects of this extraordinary effect. Among them, a big
708: controversial issue is that what is the correct definition of spin current
709: in materials with intrinsic spin-orbit coupling\cite{Murakami2, Sinova,
710: Rashba, Culcer}. As was argued in Ref.[9] and in Ref.[19], there are some
711: difficulties with the conventional definition of spin current in
712: spin-orbit-coupled systems, but it seemed that up to now there are still no
713: unanimous views about this question\cite{Murakami2, Sinova, Shen, Rashba,
714: Schliemann, Culcer, Jphu}. ( In the present paper we have used the same
715: definition of Refs.[8] ). Because no unambiguous experimental detections
716: have ever been done, on the present stage such controversial issues are
717: difficult to be clarified unambiguously by pure theoretical arguments. But
718: it could be anticipated that by combining future experimental results with
719: more accurate theoretical investigations, these controversial issues should
720: be able to be clarified unambiguously in the near future.
721:
722: \bigskip
723:
724: \textbf{Acknowledgements}
725:
726: S. Q. Shen is supported by a grant from the Research Grant Council of Hong
727: Kong. L. B. Hu is supported in part by the National Science Foundation of
728: China (Grant No.10474022).
729:
730: \begin{thebibliography}{99}
731: \bibitem{Prinz98} G.A.Prinz, Science \textbf{282}, 1660 (1998); S.A.Wolf,
732: D.D.Awschalom, R.A.Buhrman, J.M.Daughton, S.Von Molnar, M.L.Roukes,
733: A.Y.Chtchelkanova, D.M.Treger, Science \textbf{294}, 1488 (2001).
734:
735: \bibitem{Hammar99} P.R.Hammar, B.R.Bennett, M.J.Yang, M.Johnson,
736: Phys.Rev.Lett.\textbf{83}, 203 (1999).
737:
738: \bibitem{Schmidt00} G. Schmidt, D. Ferrand, L. W. Molenkamp, A. T. Filip,
739: and B. J. Van Wees, Phys. Rev. B\textit{\ }\textbf{62}, R4790 (2000).
740:
741: \bibitem{Ohno99} Y.Ohno Y, D.K.Young, B.Beschoten, F.Matsukura, H.Ohno,
742: D.D.Awschalom, Nature\textit{\ }\textbf{402}, 790 (1999).
743:
744: \bibitem{Fiederling99} R.Fiederling, M.Keim, G.Reuscher, W.Ossau, G.Schmidt,
745: A.Waag, L.W.Molenkamp, Nature \textbf{402}, 787 (1999).
746:
747: \bibitem{Mattana03} R.Mattana, J.-M. George, H. Jaffr\`{e}s, F. Nguyen Van
748: Dau, A. Fert, B. L\'{e}pine, A. Guivarc'h, and G. J\'{e}z\'{e}quel,
749: Phys.Rev.Lett.\textbf{90}, 166601(2003).
750:
751: \bibitem{Luttinger} J.M.Luttinger, Phys.Rev.\textbf{102}, 1030 (1956).
752:
753: \bibitem{Murakami03} S.Murakami, N.Nagaosa, and S.-C.Zhang, Science \textbf{%
754: 301}, 1348 (2003). ( Some supporting online material for the article was
755: given in cond-mat/0308167(unpublished)).
756:
757: \bibitem{Murakami2} S.Murakami, N.Nagaosa, and S.-C.Zhang, Phys. Rev. B
758: \textbf{69}, 235206 (2004).
759:
760: \bibitem{Hirsch99} J.E.Hirsch, Phys.Rev.Lett.\textbf{83}, 1834(1999).
761:
762: \bibitem{Zhang00} S. Zhang, Phys. Rev. Lett.\textit{\ }\textbf{85}, 393
763: (2000).
764:
765: \bibitem{Hu03} L.B.Hu, J. Gao, and S.Q. Shen, Phys. Rev. B \textbf{68},
766: 115302 (2003); $ibid$.\textbf{68}, 153303(2003).
767:
768: \bibitem{Thouless} D. J. Thouless, M. Kohmoto, M. P. Nightingale, and M. den
769: Nijs, Phys. Rev. Lett. \textbf{49}, 405(1982).
770:
771: \bibitem{Hatsugai} Y. Hatsugai, Phys. Rev. Lett. \textbf{71}, 3697(1993).
772:
773: \bibitem{Niu} G. Sundaram and Q. Niu, Phys. Rev. B \textbf{59}, 14915(1999).
774:
775: \bibitem{Sinova} J. Sinova, D. Culcer, Q. Niu, N. A. Sinitsyn, T. Jungwirth,
776: and A. H. Macdonald, Phys. Rev. Lett. \textbf{92}, 126603(2004).
777:
778: \bibitem{Shen} S.-Q.Shen, M.Ma, X.C.Xie, and F.C.Zhang, Phys. Rev. Lett.
779: \textbf{92}, 256603(2004).
780:
781: \bibitem{Schliemann} J. Schliemann and D. Loss, Phys. Rev. B \textbf{69},
782: 165315(2004).
783:
784: \bibitem{Rashba} E. I. Rashba, Phys. Rev. B \textbf{68}, (R)241315(2003).
785:
786: \bibitem{Murakami04} S.Murakami, Phys. Rev. B \textbf{69}, (R)241202(2004).
787:
788: \bibitem{Culcer} D. Culcer, J. Sinova, N. A. Sinitsyn, T. Jungwirth, A. H.
789: MacDonald, and Q. Niu, cond-mat/0309475.
790:
791: \bibitem{Sinitsyn} N.A.Sinitsyn, E.M.Hankiewicz, W.Teizer, and J.Sinova,
792: cond-mat/0310315.
793:
794: \bibitem{Jphu} J.P.Hu, B. A. Bernevig, and C. Wu, cond-mat/0310093.
795:
796: \bibitem{Inoue} J. Inoue, G. E. W. Bauer, and L. W. Molenkamp,
797: cond-mat/0402442.
798:
799: \bibitem{Dimitrova} O. V. Dimitrova, cond-mat/0405339.
800:
801: \bibitem{Burkov} A. A. Burkov and A. H. Macdonald, cond-mat/0311328.
802:
803: \bibitem{Pratt91} W.P.Pratt, Jr., S.F.Lee, J.M.Slaughter, R.Loloee,
804: P.A.Schroeder, and J.Bass, Phys.Rev.Lett.\textit{\ }\textbf{66}, 3060(1991)
805:
806: \bibitem{Gijs93} M.A.M.Gijs, S.K.J.Lenczowski, and J.B.Giesbers,
807: Phys.Rev.Lett.\textit{\ }\textbf{70}, 3343(1993).
808:
809: \bibitem{Johnson85} M. Johnson and R. H. Silsbee, Phys. Rev. Lett.\textit{\ }%
810: \textbf{55}, 1790 (1985); M. Johnson, $ibid$.\textit{\ }\textbf{70}, 2142
811: (1993); M.Johson and R.H.Silsbee, Phys. Rev. B\textbf{35}, 4959 (1987).
812:
813: \bibitem{Valet93} T. Valet and A. Fert, Phys. Rev. B\textit{\ }\textbf{48},
814: 7099 (1993).
815:
816: \bibitem{Heide} C. Heide, Phys. Rev. B \textbf{65}, 054401 (2001).
817:
818: \bibitem{Awschalom} D.D. Awschalom, D. Loss, and N. Samarth (eds.),
819: Semiconductor spintronics and quantum computation ( Springer, 2002 ).
820:
821: \bibitem{Kato} Y. Kato, R. C. Myers, A. C. Gossard, and D. D. Awshalom,
822: cond-mat/0403407.
823:
824: \bibitem{Stephens} J. Stephens, J. Berezovsky, J. P. McGuire, L. J. Sham, A.
825: C. Gossard, and D. D. Awschalom, cond-mat/0401197.
826:
827: \bibitem{Inoue2} J. Inoue, G. E. W. Bauer, and L. W. Molenkamp, Phys. Rev. B
828: \textbf{67}, 033104(2003).
829:
830: \bibitem{Edelstein} V. M. Edelstein, Sol. State. Commun. \textbf{73},
831: 233(1990).
832:
833: \bibitem{Add} When an in-plane electric field is applied in \ a
834: two-dimensional electron gas with Rashba spin-orbit coupling, the spin-orbit
835: coupling itself ( without resort to the boundary effects ) will induce a
836: homogeneous spin accumulation with the spin-polarization direction
837: perpendicular to the electric field. This effect was called the kinetic
838: magnetoelectric effect\cite{Inoue2, Edelstein}. This effect should not be
839: confused with the usual spin accumulation phenomena widely studied in the
840: literatures. Form theoretical viewpoints, the usual spin accumulation
841: phenomena are caused by the boundary effects\cite{Pratt91, Gijs93,
842: Johnson85, Valet93, Heide, Awschalom}, but the kinetic magnetoelectric
843: effect is caused by the combined action of the spin-orbit coupling and the
844: electric field and has no relation with the boundary effects.
845: \end{thebibliography}
846:
847: \end{document}
848: