【作者】 李媛；

【导师】 彭平；

【作者基本信息】 湖南大学 ， 材料科学与工程， 2022， 博士

【摘要】 均匀形核作为金属快速凝固的典型特征,对凝固组织与材料性能具有重要影响,关于其形核机制与控制方法的研究,一直是材料科学与凝聚态物理领域的一个热点课题。为了揭示快速凝固的形核特征,需要对凝固晶核（特别是临界晶核）的尺寸大小、几何构型与界面形态有一个正确认识。由于晶体的形核是一个发生在原子尺寸空间与飞秒时间尺度上的随机动态过程,导致实验上很难实现对临界晶核的原位观察和实时跟踪。为此,本文采用分子动力学模拟,结合双体分布函数、Honeycutt-Andersen（H-A）键型指数法以及团簇类型指数法等结构表征方法,系统研究了液态金属快速凝固过程中晶体团簇的形成与演化,并采用原子轨迹逆向追踪法,详细考察了不同扩展晶体团簇的结构遗传性,实现了对临界晶核的准确识别,并进一步分析了深过冷下液态金属快速凝固及其等温晶化的形核特征。主要的研究内容如下:首先,在团簇类型指数法的基础上,根据基本团簇之间不同的联结方式提出了一种可区分fcc（face-centered cubic）、hcp（hexagonal closest packed）和bcc（body-centered cubic）单晶、多晶与混晶团簇的分析方法。具体即:以交叉共享（intercross-sharing,IS）方式联结的fcc中程序称为fcc单晶团簇;除了IS联结外,还存在点共享、边共享和面共享联结的fcc扩展团簇则为fcc多晶团簇;而以IS方式联结的fcc（占主导）、hcp和bcc基本团簇则称为fcc混晶团簇。以快凝非晶Ag的等温晶化为例,本文表征了不同类型晶体团簇的内部结构与界面特征。结果表明,fcc单晶团簇的尺寸最小,多晶次之,混晶最大;fcc单晶、多晶与混晶团簇内部原子的几何构型均为非球形,但fcc单晶和多晶团簇的壳层原子中含有少量hcp和bcc原子,而混晶团簇的壳层原子则全部为非晶类原子。其次,基于液态金属快凝过程中晶体团簇具有的不同结构遗传性,发展了一种可区分fcc晶胚与临界晶核的团簇分析方法。具体即:具有连续遗传性的fcc扩展团簇称为为晶核,不具有连续遗传性的称为晶胚,而与连续遗传起始点对应的晶体团簇则称为临界晶核。采用百万级原子模拟大体系,针对液态金属Al在快速凝固过程中的均匀形核,本文详细研究了深过冷下各种fcc单晶团簇的形成与演化。通过对fcc单晶团簇连续遗传性的跟踪分析,准确识别和标度了快凝过程中先后出现的数百个临界晶核的尺寸、内部结构与界面形态。结果显示:最早出现的临界晶核含37个中心原子,几何构型偏球状。临界晶核的尺寸主要区间为10～30,平均尺寸约为26个原子,与采用等概率法计算的临界尺寸基本一致。随着形核过冷度的增加,临界晶核尺寸整体上呈减小趋势,形核率则先增后减。此外,同一温度下的临界晶核不仅尺寸不同,内部结构与界面形态也各不相同,但大部分临界晶核是非球状的不规则几何构型,其液固界面则为fcc/hcp-液态原子多相结构。最后,模拟研究了快凝金属Al在不同温度下等温晶化的形核过程,探究了过冷液态金属Al的均匀形核极限,并分析了其动力学和热力学特征。基于对临界晶核的识别确定了均匀形核极限温度,并探究了Stokes-Einstein关系的破裂对均匀形核极限的影响。结果表明:液态金属Al的均匀形核极限温度_ks=0.51_m（_m为Al的理论熔点）,形核能垒消失时对应的热力学旋节点温度略低于_ks,为0.45_m。当体系达到均匀形核极限时,其热力学形核能垒为非零的极小值,体系中出现了动力学临界慢化现象。此外,Stokes-Einstein关系的破裂能有效延长结构驰豫时间,但并不是体系达到均匀形核极限的必要条件。最后通过可视化分析揭示了液态金属Al在均匀形核极限附近的凝固形核特征。本文严格区分了单晶、多晶和混晶团簇,完善了对纳米尺寸的晶胚和晶核的结构表征;并提出了一种识别临界晶核的团簇分析方法,该方法能准确识别大体系或深过冷下同时形成的多个临界晶核,弥补了前人只能通过宏观热力学参数判定小体系中单个临界晶核的不足;实现了对快速凝固和等温晶化过程大量临界晶核的统计分析,进而揭示了临界晶核的尺寸大小、几何构型和界面形态等特征,该研究有助于人们对深过冷下的临界晶核和形核特征有一个更全面、清晰的认识。更多还原

【Abstract】 Homogeneous nucleation,as the typical feature of rapid solidification of metals and alloys,has a significant influence on the final solidified structure and the properties of materials.Therefore,the nucleation mechanism and its control methods have always been a interesting topic in the research fields of materials science and condensed matter physics.To reveal the nucleation features of the rapid solidification,the foremost task is to have a comprehensive and accurate understanding on the size,geometry and interface morphology of the critical nucleus.As the nucleation is a stochastic dynamic process which involves exceedingly small time and length scales,molecular dynamics（MD）simulation can provide a unique insight into the microscopic aspects of nucleation,while it is difficult to directly obser ve and track the formation of critical nucleus experimentally.In this thesis,we investigate the nucleation process of liquid metals by MD simulations.By means of structural analysis methods such as pair distribution function,H-A（Honeycutt-Andersen）bond-type index method and CTIM（cluster-type index method）,the evolutions of microstructures of the simulated systems during crystallization are analyzed.By the reverse tracing of atomic trajectory,the critical nuclei are identified based on the structural heredity of different clusters.Further,the characteristics of nucleation of the metals are revealed under rapid solidification and deep supercooling.Firstly,on the basis of CTIM,a structural analysis method is proposed to distinguish fcc（face-centered cubic）,hcp（hexagonal closest packed）and bcc（body-centered cubic）single-crystal,poly-crystal and hybrid-crystal clusters according to the linkage modes among basic clusters.The fcc medium-range-order,which is composed of fcc basic clusters linked by intercross-sharing（IS）modes,is defined as an fcc single-crystal cluster.The fcc extended cluster,which is composed of fcc basic clusters linked by vertex-sharing（VS）,edge-sharing（ES）or face-sharing（FS）besides IS modes,is defined as an fcc poly-crystal cluster.The fcc extended cluster,which is composed of fcc（dominant）,hcp and bcc basic clusters linked by IS modes is defined as an fcc hybrid-crystal cluster.The internal structures and interface characteristics of different types of extended crystal clusters are characterized and illustrated in an example of the isothermal crystallization of quenched amorphous Ag.The simulation results show that the size of fcc single-crystal cluster is the smallest,followed by fcc poly-crystal cluster,and fcc hybrid-crystal cluster are the largest.The geometric configurations of internal atoms of the fcc single-crystal,poly-crystal and hybrid-crystal clusters are all non-spherical shape.The shell atoms of fcc single-crystal and poly-crystal cluster contain a small amount of hcp and bcc atoms,while the shell atoms of hybrid-crystal cluster are all amorphous and liquid atoms.Further,a unique method to distinguish critical nuclei from embryos is proposed based on the structural heredity of crystal clusters.The extended crystal cluster which has continuous heredity is defined as a nucleus,while the one which has no continuous heredity is called as an embryo,and the nucleus corresponding to the onset point of continuous heredity is identified as a critical nucleus.The rapid solidification process of millions-atom of metal Al is simulated,and the formation and evolution of fcc single-crystal clusters is studied.Hundreds of critical nuclei are further identified by analyzing the continuous heredity of fcc single-crystal clusters during rapid solidification,and the crucial information such as critical size,geometric configuration and interfacial structure of nuclei are directly obtained.The first critical nucleus contains 37 atoms,which have a spherical-like shape.The main size distribution range of critical nuclei is 10～30,and the average size of critical nuclei is about 26,which is completely consistent with the critical size calculated by the equal-probability method.With the increase of nucleation supercooling,the size of the critical nuclei tends to decrease as a whole,and the nucleation rate first increases and then decreases.Moreover,different critical nuclei at same temperature are not only different in size,but also in internal structure and interface morphology.Most of the critical nuclei have non-spherical and irregular geometrical configurations,and the interfaces of most of nuclei are fcc/hcp-liquid structures.Finally,the homogeneous nucleation limit（HNL）of liquid Al under deep supercooling is studied and its kineti c and thermodynamic characteristics are revealed.The nucleation processes of supercooled liquid Al at different temperatures are simulated,and the HNL is deduced out based on the identification of critical nuclei.The influence of the breakdown of Stokes-Einstein（SE）relation on the HNL is also investigated.The temperature_ks=0.51_m（_m is the theoretical melting temperature of Al）corresponding to the HNL of Al is determined,and the thermodynamic spinodal temperature_ts corresponding to the vanishing of nucleation barrier is found to be0.45_m,which is slightly lower than_ks.It is revealed that the thermodynamic nucleation barrier at_ks is very small but non-zero,and a kinetic critical slowing down phenomenon is found near_ks.Additionally,it is found that the breakdown of SE relation can prolong the structural relaxation time and then promote the occurrence of the HNL,but the breakdown is not a prerequisite for the occurrence of the HNL.At last,the nucleation characteristics of supercooled liquid Al at the HNL are intuitively observed by means of visualization.In this thesis,single-crystal,poly-crystal and hybrid-crystal clusters are strictly distinguished,and the structural characterization of embryos and nuclei wit h nanometer dimensions is improved.A cluster analysis method to identify critical nuclei is proposed,which can accurately identify multiple critical nuclei formed at the same time in large systems or under deep undercooling.Statistical analysis of hundreds of nuclei reveals the size,geometry and interface morphology of the critical nuclei,which is helpful for researchers to have a more comprehensive and clear understanding of critical nuclei and nucleation characteristics under deep supercooling.更多还原