【作者】 蒋建华；

【导师】 吴明卫；

【作者基本信息】 中国科学技术大学 ， 凝聚态物理， 2010， 博士

【摘要】 半导体自旋电子学的目标是通过利用载流子的自旋自由度开发出速度更快、能耗更低的电子学器件。这一目标要求能产生、操控、维持、传输和读出载流子自旋极化。其中至少操控、维持和传输载流子自旋极化都和自旋动力学相关,因而自旋动力学成为半导体自旋电子学的中心内容之一。自旋动力学背后丰富的物理也是导致它受到人们广泛地关注的原因。通常,自旋动力学包括相干部分和耗散部分。相干的自旋运动通常很简单而平凡,而耗散的自旋运动则往往很复杂而有趣。很多实际的自旋现象的复杂性和丰富性往往根源于耗散动力学。本论文集中讨论Ⅲ-Ⅴ族半导体(它是半导体自旋电子学中被广泛研究的材料)及其纳米结构中的自旋动力学。特别地,我们关注Ⅲ-Ⅴ族半导体及其纳米结构中的自旋弛豫和自旋去相位。在第一章中我们首先简要地介绍半导体自旋电子学的背景。具体而言,我们介绍了它的历史、目标、已有的成就和挑战。我们也介绍了一些典型的自旋电子学器件的构想、在半导体中产生和探测自旋极化的方法、Ⅲ-Ⅴ族半导体中的自旋相互作用。我们还介绍了自旋弛豫和自旋去相位的概念并基于一些简单的模型对自旋弛豫和自旋去相位进行了一些一般性的讨论,并介绍Ⅲ-Ⅴ族半导体中自旋弛豫和自旋去相位的主要机制。之后在第二章我们回顾了以往文献中对Ⅲ-Ⅴ族半导体中自旋弛豫和去相位的研究,包括实验和单体理论的研究。我们既回顾了金属区域的研究,也回顾了绝缘体区域的研究。我们还回顾了Ⅲ-Ⅴ族磁性半导体中的自旋弛豫的实验研究。值得指出的是,之前文献中的理论绝大部分都是在单体理论的框架下。而单体理论存在很多的问题,很多有趣而重要的结果只有在多体理论的框架下才能得到。我们在这一章中既指出了单体理论的问题,也通过实验结果提及了一些多体理论才能解释和理解的有趣现象。在第三章中我们简要地介绍了动力学自旋Bloch方程方法。它是我们后续研究的基础。动力学自旋Bloch方程用全微观的方式包含了描述半导体中自旋动力学的所有要素：自旋相互作用导致的自旋相干进动和所有相关的散射。此外,它还仔细地处理了屏蔽效应对散射和库仑Hartree-Fock项的影响。我们分别介绍了本征型量子阱中考虑自旋和光学关联的动力学自旋Bloch方程和n型量子阱中仅考虑自旋关联的动力学自旋Bloch方程。在此基础上,在第四章,我们详细地阐述对体材料Ⅲ-Ⅴ族半导体中的电子自旋弛豫的系统研究。我们的研究覆盖了n型、本征型和p型Ⅲ-Ⅴ族半导体。我们主要的发现是：在简并区和非简并区自旋弛豫时间随各种条件的变化通常在定性上都是不一样的,要理解自旋弛豫时间随各种条件的变化,首先要判断电子和空穴系统是处于简并区还是非简并区。另外我们发现由于屏蔽和Pauli阻塞等因素的影响,散射随各种物理条件的变化很丰富,自旋弛豫(特别是D’yakonov-Perel’自旋弛豫)的行为往往比人们所理解的要丰富、复杂得多。具体而言,我们得到了如下重要的结论：在n型、本征型和大部分的p型Ⅲ-Ⅴ族半导体中Elliott-Yafet机制不重要；在n型和本征型半导体中,由于从非简并区到简并区过渡,D’yakonov-Perel’自旋弛豫时间随电子浓度的增大而先增大后减小,自旋弛豫时间在TF～T(TF是费米温度,T是体系温度)附近出现一个峰；在n型半导体中不同浓度下自旋弛豫时间随温度的变化不一样：在低浓度下施加应变的情形下自旋弛豫时间可能出现非单调的温度变化,而在无应变的情形下自旋弛豫时间随温度上升而单调变短；在高浓度下,自旋弛豫时间随温度的上升而延长；在常见的一些Ⅲ-Ⅴ族半导体如GaAs、GaSb和InSb等的本征型体材料中,Bir-Aronov-Pikus机制不重要；当电子系统或空穴系统处于简并区时,Pauli阻塞对Bir-Aronov-Pikus自旋弛豫有很强的抑制作用；本征型半导体中,小极化下,D’yakonov-Perel’机制导致的自旋弛豫时间随温度增大先增大后减小,出现一个峰,峰的位置出现在T(?)TF／3附近；Kp型Ⅲ-Ⅴ族半导体中,在高激发浓度下,由于屏蔽的作用,D’yakonov-Perel’自旋弛豫时间随温度升高先增大后减小,出现一个峰,峰的位置出现在电子的费米温度附近T～TF; D’yakonov-Perel’自旋弛豫时间则随空穴浓度的增大出现先增大后减小再增大的奇特行为；最后,在n型半导体中,强电场导致电子自旋弛豫时间减小,且电子迁移率越高电场的影响越大。我们透彻地阐释了以上现象背后的物理,并揭示出库仑散射在自旋弛豫中所扮演的重要角色。尽管我们的研究集中在闪锌矿结构的Ⅲ-Ⅴ族半导体中,我们得到的上述结论中有很多是普适性的。它们可以扩展到非闪锌矿结构、Ⅱ-Ⅵ族半导体,甚至是其它的结构类似的半导体(只要存在自旋轨道耦合),低维纳米体系等等。特别值得一提的是,我们的一些预言已经被最近的一些实验所证实[1-7]。第五章给出了我们和德国Regensburg大学Schuller实验组合作对(001)GaAs量子阱中自旋弛豫的各向异性的研究。实验组测量了低温下高迁移率的GaAs量子阱中自旋弛豫的磁各向异性。发现自旋弛豫的各向异性可以通过磁场显著地调节。特别地,当磁场沿[110]方向时,自旋弛豫时间在B=0.2 T处出现了一个谷；而当磁场沿[110]方向时,自旋弛豫时间在B=0.5 T处出现了一个峰。所观测到的现象无法用原来的单体理论解释。我们通过基于全微观的动力学自旋Bloch方程的计算和实验结果符合得很好。我们对其中的物理进行了解释。进一步地,我们预言该样品中[110]方向的自旋弛豫时间可以达到几个纳秒,比[110]方向的自旋弛豫时间大两个量级以上。以上发现对半导体自旋电子学中调控自旋弛豫具有很大的意义。此外,第六章回顾我们基于全微观的动力学自旋Bloch方程方法对稀磁半导体顺磁性GaMnAs量子阱中的电子自旋弛豫的研究。我们既研究了Mn占据填隙位置的n型GaMnAs量子阱,也研究了Mn主要替换Ga的p型GaMnAs量子阱。对于n型GaMnAs量子阱,我们发现,自旋弛豫完全由D’yakonov-Perel’机制占主导。一个显著的结果是,我们发现自旋弛豫时间随Mn的参杂浓度的变化出现了一个峰。这个峰是由于电子系统处于简并区和非简并区时动量散射和自旋进动的非均匀扩展随Mn的参杂浓度的变化不同而导致自旋弛豫时间随Mn的参杂浓度变化的趋势不同。有趣的是,在p型GaMnAs量子阱中自旋弛豫时间随Mn的参杂变化也有一个峰。这是D’yakonov-Perel’自旋弛豫机制和s-d-Elliott-Yafet和Bir-Aronov-Pikus机制竞争的结果。我们计算得到的自旋弛豫时间的峰的位置和Awschalom实验组实验测量[8]得到的很一致。另外,我们还确定了各种条件下占主导的自旋弛豫机制。这为理论和实验研究提供了有用的信息。我们还系统地研究了各种温度、光激发浓度、磁场下的自旋弛豫,给出了其背后的物理。我们得到的结论和实验结果一致[8-11]。之后,在第七章我们介绍含时系统的动力学和处理含时系统的理论方法。此章为我们对强THz场驱动下的自旋动力学的研究提供一个背景和相关理论的介绍。我们首先简要介绍了凝聚态物理中的含时驱动系统。我们还介绍强THz电磁场相关的技术和物理,简要回顾了强THz场对半导体光学性质和输运性质的影响。然后我们回顾了无耗散下的含时驱动系统的动力学,介绍了求解含时Schrodinger方程的Floquet-Fourier方法,简要讨论了Floquet波函数的性质。然后我们介绍了处理含时驱动系统耗散动力学的Floquet-Markov理论。在此基础上,在第八章阐述我们对强THz场驱动的量子点中单个电子的自旋动力学的研究。我们首先得到强THz场驱动下Schrodinger方程的严格解,并研究了强THz对电子态密度的影响。我们发现在自旋轨道耦合存在时,强THz电场可以操控电子自旋,并在量子点中诱导出自旋极化,同时对电子态密度有很大的影响。在此基础上,我们考虑耗散,加入导致自旋弛豫的电子-声子散射。我们的研究结果表明,强THz场能极大的影响自旋弛豫。特别地,在强THz磁场下,sideband效应强烈地调制了自旋翻转的电子-声子散射的速率,极大地改变了自旋弛豫时间。然后在第九章给出我们对强THz场驱动下的多电子系统的自旋动力学的研究。我们考虑的是InAs量子阱中的二维电子系统。通过Floquet-Markov理论和非平衡Green函数方法,我们首先建立了强THz场驱动下量子阱中二维电子系统的动力学自旋Bloch方程。在该方程中,我们对THz场进行了非微扰的处理,在散射中包含了sideband效应,并在对散射的处理上超越了旋波近似。我们包含了所有相关的散射：电子-杂质、电子-声子和电子-电子散射。我们的方法具有很大的普遍性,可以扩展到任意的强周期场驱动下具有任意自旋轨道耦合的多载流子系统。通过数值求解动力学自旋Bloch方程,我们研究了强THz场对InAs量子阱中二维电子系统的自旋动力学的影响。其中,我们主要讨论了THz场产生的稳态自旋极化和THz场对自旋弛豫的影响。我们发现THz场诱导出一个稳态的自旋极化。尽管这极化是Cheng和Wu在没有考虑耗散的时候最早预言的[12],我们发现在耗散存在时它仍然不为零,而且它的值可以达到很大(例如,7%)。这表明强THz场是产生自旋极化的很有效的手段。我们的研究表明稳态自旋极化是由THz导致的有效磁场诱导的。它包含两个部分的贡献,一个直接来自自旋轨道耦合,另一个来自THz场产生的电流和自旋轨道耦合的合作效应。由于我们对散射的处理超越了旋波近似,我们发现了很多有趣的特征,这些特征在对散射做旋波近似下是不会出现的。第一个特征是,稳态自旋极化总是相对于激发它的有效磁场有一个推迟。另外一个特征是,THz场导致的电流会产生一个有效磁场。我们发现,在对散射做旋波近似下：稳态自旋极化和激发它的有效磁场间没有推迟。此外,更重要的是,在对散射做旋波近似下,散射会保持kx→-kx对称性,从而无法得到THz场导致的电流。我们研究了稳态自旋极化幅度随THz场的场强和频率的变化。我们发现THz场对自旋动力学的两个主要的影响是：(1)热电子效应和(2)THz场导致的有效磁场。这两个效应都随THz场的场强的增大或频率的减小而增强。其中THz场导致的有效磁场是诱导出电子系统稳态自旋极化的物理根源。由于THz场导致的有效磁场随THz场的场强的增大或频率的减小而增强,稳态自旋极化的幅度因而增大。但是在THz场的场强较大或频率较小时,热电子效应导致电子温度显著地升高,从而导致稳态自旋极化随THz场的场强的增大或频率的减小而下降。我们发现强THz场对自旋弛豫也有很大的影响。它的影响也来自上面所说的(1)热电子效应和(2)THz场导致的有效磁场。我们发现,在杂质浓度很小的情况下,热电子效应和THz场导致的有效磁场相互竞争,导致自旋弛豫时间随THz场的场强的增强或频率的减小出现先增大再减小的结果。在杂质浓度较大时,热电子效应得到增强完全超过了THz场导致的有效磁场的影响,使得自旋弛豫时间随THz场的场强的增强或频率的减小而单调地减小。最后在第十章详细地回顾了我们基于运动方程方法对GaAs量子点中各种机制导致的自旋弛豫时间和自旋去相位时间的系统研究。在此研究中我们考虑了自旋轨道耦合和电子-声子散射共同作用机制、直接的自旋-声子耦合机制、声子导致的g因子涨落机制、超精细相互作用机制以及超精细相互作用和电子-声子相互作用的共同作用机制。我们在各种条件下比较了这些机制对自旋弛豫和自旋去相位的贡献。研究了它们随各种物理条件变化的规律并揭示了其中的物理。我们发现GaAs量子点中的自旋弛豫和自旋去相位不是受到单一的自旋退相干机制的影响的,各个机制主导自旋弛豫和去相位的范围不一样。在某些情况下,有可能几种机制都很重要。我们的计算和实验符合得很好[13]。我们还给出了费米黄金规则方法的适用范围。我们发现的自旋去相位时间随温度变化的规律已经被实验所证实[14]。我们的研究对于理解GaAs量子点中的自旋弛豫和自旋去相位,进一步地,对操控自旋退相干以及基于此的量子计算和量子信息处理,具有很重要的意义。更多还原

【Abstract】 The aim for a new generation of electronics with much higher operation speed and lower power dissipation via exploiting the carrier spin degree of freedom motivates the studies in the field "semiconductor spintronics". Spin dynamics is one of the central issues in semiconductor spintronics, not only due to its relevance in spin-based devices, but also due to its rich phenomena and physics. In most cases spin dynamics consists of the coherent part and the dissipative part. The coherent part is usually simple and trivial, whereas the dissipative part is complicated and interesting. The complexity and variety of genuine spin phenomena usually roots in the dissipative dynamics. In this dissertation we focus on spin dynamics in III-V semiconductors and their nanostructures which are the widely studied materials in semiconductor spintronics community. Especially, we focus on spin relaxation and spin dephasing in III-V semiconductors and their nanostructures.We first briefly introduce the field of semiconductor spintronics in Chapter I by ex-ploring its history, aims, achievements and challenges. We then introduce some prototype spintronic devices, which historically stimulated the community a lot. We also introduce the methods of generating and detecting spin polarization. We then devote a lot of para-graphs to spin interactions in III-V semiconductors, which is the physical origin of all spin phenomena. After that we introduce the relevant spin relaxation mechanisms in III-V semiconductors.After that, we comprehensively review the literature of experimental studies and single-particle theories on spin relaxation in III-V semiconductors in Chapter II. We review spin relaxation in both metallic and insulating regimes. We also review spin relaxation in paramagnetic (III,Mn)Ⅴsemiconductors. It should be pointed out that most of the previous theoretical studies are based on the single-particle approach, which fails in many cases. Many interesting features can only be obtained by the many-body theory. We also mention some experimental results which can only be understood in the many-body framework, besides explicitly pointing out the problems in the single-particle theories.In Chapter III we briefly introduce the kinetic spin Bloch equation approach, which is the fundamental of our work on spin dynamics in semiconductors. Kinetic spin Bloch equations describe spin dynamics in a fully microscopic fashion:they include both spin precession due to all spin interactions and all relevant scatterings. Furthermore, screening is treated carefully as it will affect both scattering and the Coulomb Hartree-Fock term. We introduce the kinetic spin Bloch equations including the optical and spin correlations in intrinsic quantum wells as well as the kinetic spin Bloch equations with only the spin correlations in n-type quantum wells as examples.After that, we elaborate on our comprehensive research on electron spin relaxation in bulk III-V semiconductors based on kinetic spin Bloch equation approach in Chapter IV. Our studies cover n-type, intrinsic and p-type III-V semiconductors. We find that the dependences of spin relaxation time in degenerate regime is qualitatively different from that in non-degenerate regime. To understand the various dependences of the spin relaxation time, the first job is to determine in which regime the electron/hole system is. Besides, we find that, due to factors such as screening and Pauli blocking, depen-dences of momentum scattering rates are complicated, which lead to intricate and various behaviors in spin relaxation. Especially the behavior of the D’yakonov-Perel’spin relax-ation time is much more complicated than what was understood in the literature. Our main findings are:In n-type, intrinsic and most of the p-type semiconductors, the Elliott-Yafet mechanism is less important than the D’yakonov-Perel’mechanism, even for the narrow band-gap semiconductors such as InSb and InAs; Due to the crossover from non-degenerate regime to degenerate regime, the density dependence of spin relaxation time exists a peak in the metallic regime around TF～T in both n-type and intrinsic materi-als; In n-type III-V semiconductors, the temperature dependence of spin relaxation time varies for different densities; Specifically, in low-density case with strain, the temperature dependence can be nonmonotonic, whereas in the case without strain the spin lifetime de-creases with increasing temperature; In high density case, spin lifetime can increase with increasing temperature; In common intrinsic III-V semiconductors such as GaAs, GaSb and InSb, the Bir-Aronov-Pikus mechanism is found to be negligible compared with the D’yakonov-Perel’one; In the case of small initial spin polarization, there is a peak in the temperature dependence of spin relaxation time located around T～TF/3, which is due to the nonmonotonic temperature dependence of the electron-electron Coulomb scatter-ing in intrinsic semiconductors; In p-type semiconductors, under high excitation density, due to the screening effect, the D’yakonov-Perel’spin relaxation time first increases then decreases with increasing temperature with a peak around T～TF; The D’yakonov-Perel’ spin relaxation time exhibits intriguing behaviors in hole density dependence—it first in-creases then decreases, and then increases with increasing hole density; Finally, in n-type semiconductors, high electric field leads to shorter spin lifetime and the effect of electric field increases with increasing mobility. We elaborate thoroughly on the underlying physics of the above intriguing behaviors. We point out that some of the behaviors are universal which also exist in low-dimensional structures,Ⅱ-VI semiconductors and Wurtzite semi-conductors, or other materials with similar structures (given that the spin-orbit coupling is finite). It should be mentioned that some of our predictions have been confirmed by recent experiments [1-7].We then describe our investigation (in collaboration with experimentalists in the Schuller group in Regensburg University) on anisotropic spin relaxation in (001) GaAs quantum wells in Chapter V. The experimentalists measured the magneto-anisotropy of spin relaxation in high mobility GaAs quantum wells. They found that the anisotropy in spin relaxation can be tuned largely by the magnetic field. Specifically, when the magnetic field is along the [110] direction, spin relaxation time exists a valley at B= 0.2 T in the magnetic field dependence, whereas there is a peak when the magnetic field is along [110] direction at B= 0.5 T. The observed phenomena can not be explained by previous theory. We employed the kinetic spin Bloch equation approach to calculate spin relaxation time under the experimental conditions. We find good agreement with the experimental data. We explored the underlying physics. Moreover, we predicted the lifetime for the spin com-ponent along the [110] direction to be several nanoseconds—two orders larger than that of the spin component along [110]. These findings are valuable to spin lifetime manipulation in semiconductor spintronics.We present our systematic studies on electron spin relaxation in paramagnetic GaM-nAs quantum wells based on kinetic spin Bloch equation approach in Chapter VI. We study the spin relaxation in both the n-type GaMnAs quantum wells where most of Mn ions take the interstitial positions as well as the p-type GaMnAs quantum wells where most of Mn ions substitute Ga atoms. For n-type GaMnAs quantum wells, we find that spin relaxation is completely dominated by the D’yakonov-Perel’mechanism. Remark-ably, the Mn concentration dependence of the spin relaxation time is nonmonotonic and exhibits a peak. This is due to the fact that the momentum scattering and the inhomoge-neous broadening have different density dependences in the non-degenerate and degenerate regimes. Interestingly, in p-type GaMnAs quantum wells, there also exists a peak in the Mn concentration dependence. Differently, this peak is due to the competition between the D’yakonov-Perel’mechanism and the other mechanisms such as the s-d exchange, Elliott-Yafet and Bir-Aronov-Pikus mechanisms. We reproduced the peak position measured by Awschalom group [8]. Moreover, we determine the dominant spin relaxation mechanisms in various regimes, which offer very important information for further studies. The tem-perature, photo-excitation density and magnetic field dependences of the spin relaxation time are investigated systematically with the underlying physics revealed. Our results are consistent with the recent experimental findings [8-11].In Chapter VII, we review kinetics in driven time-dependent system, including the theoretical methods. This chapter serves as a background introduction for our studies in spin dynamics under intense THz driving field. We first briefly introduce several driven time-dependent systems in condensed matter physics. We then introduce THz technology and related physics. We briefly review the effects of intense THz field on the transport and optics of semiconductors. We then review the kinetics in driven time-dependent system in the dissipation-free limit. We introduce the Floquet-Fourier approach to time-dependent Schrodinger equation and discuss briefly the properties of the Floquet wavefunction. After that, we present the the Floquet-Markov theory for the dissipative kinetics in driven time-dependent system.After that we review our studies on spin dynamics in quantum dots under intense THz field in Chapter VIII. We first obtain the exact solution to the Schrodinger equations, and then study the effect of intense THz field on the density of states. We show that in the presence of spin-orbit coupling the THz electric field can be used to manipulate spin and induce a spin polarization perpendicular to the electric field. After that we include the electron-phonon scattering and investigate the effect of intense THz field on spin relaxation. We find that intense THz magnetic field can strongly affect the spin relaxation via the sideband modulated spin-flip electron-phonon scattering.In Chapter IX, we present our investigations on spin kinetics in many-electron system under intense THz driving field. We consider the two-dimensional system confined in an InAs quantum well. We first construct the kinetic spin Bloch equation via Floquet-Markov theory and nonequilibrium Green function theory. In the kinetic spin Bloch equations, we treat the THz field nonperturbatively, where all the sidebands are included. We treat the scattering beyond the rotating wave approximation. We include all relevant scatterings, including the electron-impurity, electron-phonon and electron-electron scatterings. Our approach is quite general and can be applied to other many-carrier system with arbitrary spin-orbit coupling. By numerically solving the kinetic spin Bloch equations, we study the effect of intense THz laser field on spin kinetics. Specifically, we mainly discuss the THz field-induced spin polarization and the effect of THz field on spin relaxation. We find that the THz field induces a steady-state spin polarization which first predicted by Cheng and Wu [12] in dissipation free case, still exists in the presence of scattering. The steady-state spin polarization can be as large as 7%, which indicates that intense THz field is an efficient tool for generating spin polarization. Our research reveals that there are two physical origins of the induced spin polarization. The first part is induced by the effective magnetic field directly due to the spin-orbit coupling. The second part is induced by the effective magnetic field due to the combined effect of THz field-induced current and spin-orbit coupling. As we treating the scattering beyond the rotating wave approximation, we find many interesting features, which are absent within the rotating wave approximation. The first feature is that there is always a retardation of the spin polarization in response to the THz field-induced effective magnetic field. Another is that the THz field can induce a current which leads to an effective magnetic field in the presence of spin-orbit coupling. If the scattering is treated in the rotating wave approximation, there is no retardation. More importantly, within such approximation the scattering will keep the kx→-kx symmetry, and there is no THz-field-induced current. We investigate the dependence of the amplitude of the steady-state spin polarization on the strength and frequency of THz field. We find that the main factors affecting the amplitude of spin polarization are that:(1) THz field-induced hot-electron effect, (2) the THz field-induced effective magnetic field. Both effects increase with increasing THz strength or decreasing frequency. As the spin polarization is induced by the effective magnetic field, higher effective magnetic field leads to higher spin polarization. However, the hot-electron effect reduces the spin polarization. We find that the amplitude of the induced spin polarization first increases then decreases with increasing THz field strength or decreasing THz frequency. The decrease is due to that the prominent hot-electron effect at high THz field or low THz frequency. We also find that the intense THz field strongly affects the spin relaxation. The effect also comes from (1) the hot-electron effect and (2) the THz field-induced effective magnetic field. We find that in the case with very small impurity density, the two factors compete to each other and lead to the nonmonotonic dependence of spin relaxation time with increasing THz field strength or decreasing frequency. At high impurity density the hot-electron effect dominates and the spin relaxation time decreases monotonically with increasing THz field strength or decreasing frequency.We review our systematic research on spin relaxation and dephasing in GaAs quantum dots in Chapter X. In our investigation we include various spin decoherence mechanisms, such as the hyperfine interaction, spin-orbit coupling together with the electron-phonon scattering, the g-factor fluctuations, the direct spin-phonon coupling due to the phonon-induced strain, and the coaction of the electron-phonon interaction together with the hyperfine interaction. The relative contributions to spin relaxation and dephasing from these mechanisms are compared under various conditions, where their dependences are studied with the underlying physics revealed. We find that both spin relaxation and spin dephasing are not determined by one spin decoherence mechanism solely. Each mechanism dominates one regime. In some situations, several mechanisms are comparable. Our calculation agrees well with experiments [13]. We also give the parameter regimes where the Fermi Golden rule is valid. The predicted temperature dependence of spin dephasing time has been confirmed by recent experiment [14]. Our study is valuable for the understanding of spin relaxation and dephasing in GaAs quantum dots, which serves as the ground for further studies on the manipulation of spin decoherence and for quantum information processing.更多还原