【作者】 马文；

【导师】 经福谦； 祝文军；

【作者基本信息】 国防科学技术大学 ， 物理学， 2011， 博士

【摘要】 冲击压缩下金属材料的塑性变形及相变机制一直以来都是材料学和冲击波物理研究的重点和难点。本文以纳米多晶延性金属铝、铜和铁为研究对象,使用分子动力学模拟为研究手段研究了面心立方结构的铝和铜的微观塑性变形及体心立方结构的铁的冲击相变过程。由于冲击波阵面是冲击压缩下材料塑性变形和相变中十分重要的宏观表现,所以在研究中重点关注了冲击波阵面的详细结构、影响冲击波阵面结构的因素和冲击波阵面与微观塑性变形或相变过程的对应关系。首先,研究了构建分子动力学模拟所需的纳米多晶金属样本的方法。构建样本包括两个步骤:一是用Voronoi几何方法生成纳米多晶样本的初始结构,二是对初始的几何构型进行弛豫以降低由非物理晶界带来的高能量和高应力。在此过程中我们首次使用了样本的残余内应力来判定纳米多晶样本是否与实验制备的纳米多晶金属样品相一致。弛豫纳米多晶样本时,可首先快速冷凝Voronoi方法生成的初始样本10ps左右使之达到局域最低能量状态,更长时间的快速冷凝对降低能量和内应力没有明显的效果。之后样本可在恒温零应力周围环境下模拟退火40～100ps后达到最低能量和最小残余内应力状态,选取的温度越高则退火时间可设置的越短,但退火温度不能高于材料熔点的0.65倍,更长时间的退火也没有必要。同时,我们在弛豫过程中监测了晶界结构、体系平均内应力和能量的总体和局域变化情况,并计算了不同弛豫条件下终态样本的弹性常数。研究结果表明终态样本的能量、残余内应力和弹性常数都接近实验制备的纳米多晶金属。其次,定性研究了晶界对纳米多晶铝的冲击波阵面结构及塑性变形机制的影响,定量研究了纳米多晶铜的冲击波阵面宽度与冲击压缩条件的关系。通过观察冲击应力和粒子速度剖面与塑性变形过程的原子尺度图像获得了晶界相关的塑性变形对冲击波阵面的影响。通过计算一系列压缩条件下的冲击波阵面宽度粗略得到了冲击应力与波阵面宽度的定量关系,与他人的定量关系符合的比较好。研究结果表明:在弹性先驱波之后,是晶界的滑移和变形主导了前期的塑性变形机制;然后是不全位错在晶界上成核向晶粒内传播,并在晶粒内形成堆垛层错、孪晶和全位错的过程主导了后期的塑性变形机制。冲击波阵面扫过之后留下的结构特征是堆垛层错和孪晶留在晶粒内,大部分全位错则湮灭于对面晶界。同时发现对纳米多晶铝而言,晶界相关的塑性变形对冲击波阵面的贡献与位错相关的塑性变形的贡献是可以比拟的。然后,比较研究了具有相同织构和晶粒度的纳米多晶铝和铜的塑性变形机制及冲击波阵面结构。通过比较铝和铜在相同应力或应变的冲击压缩下的一维、二维冲击波剖面和塑性变形过程的原子图像发现:纳米多晶铜在弹性变形阶段的速率快于纳米多晶铝,其原因是铝的晶格常数大于铜导致需要更长时间压缩来使其发生塑性变形;晶界主导的塑性变形的持续时间和对冲击波阵面宽度的贡献小于纳米多晶铝,整体的冲击波阵面宽度小于纳米多晶铝的宽度,塑性变形后的位错密度高于纳米多晶铝。纳米多晶铝和铜的塑性变形机制也略有不同,在纳米多晶铝中观察到了不全位错,全位错和形变孪晶,而在纳米多晶铜中却只有不全位错。另一方面,这两种纳米多晶金属的晶界都会发生滑移和增厚等晶界主导的塑性变形,但是我们还在纳米多晶铝的塑性变形过程中发现了晶界的弯曲,而在纳米多晶铜中却没有。产生这些塑性变形机制差异的主要原因是诸如晶格常数、层错能和剪切模量等材料本征参数使得铝中发射位错的临界分切应力高于铜,从而使纳米多晶铝中晶界相关的塑性变形持续时间更长,产生的位错少,晶界相关的变形机制更多,并最终使冲击波阵面更宽。最后,研究了不同晶粒度的纳米多晶铁在冲击压缩下的结构相变过程。研究显示纳米多晶铁的冲击结构相变发生的临界冲击应力在15 GPa左右。纳米多晶铁在经过弹性压缩变形后,晶界导致的塑性变形开始发生,然后大多数相变从晶界成核并最终发展为大规模相变。不同变形过程在应力和粒子速度剖面上能得到清晰的体现,并通过微观原子结构分析分辨。冲击压缩后除了晶界处的铁原子外大部分发生了结构相变,最终的微观结构以晶界原子和六角密排结构原子为主,少量面心立方结构原子充当了六角密排结构新相的孪晶界。在所研究的晶粒度范围内,晶粒越大包含的晶界缺陷原子数越少并且最终的相变比例越高,消耗的能量更多并进一步使得终态的冲击应力越低。本文初步探讨了晶界这种材料中普遍存在的缺陷对冲击压缩下延性金属的塑性变形及相变机制的影响,并如何进一步影响冲击波阵面这一宏观表现。研究结果有助于深刻理解材料的塑性变形及相变机制,并对揭示纳秒-微米尺度的宏观-微观过程之间的联系和构建微观过程的物理图像具有参考意义。虽然一些定量结果与传统粗晶材料的冲击压缩实验结果存在差异,但本文获得的认识对于一般的粗晶材料动态响应机制研究也有帮助,例如在解读冲击波阵面结构的微观机理上。更多还原

【Abstract】 The mechanisms of plastic deformation and phase transformation of metals under shock compression have been important and difficult subjects in material sciences and shock wave physics for a long time. In this paper, we have studied the microscopic mechanisms of plastic deformation of the face-centered cubic (fcc) structured nanocrystalline (nc) aluminum and copper, and phase transformation of body-centered cubic (bcc) structured nc iron under shock compression by means of molecular dynamics (MD) simulations. Because the shock front is one of the most important macroscopic representations of the plastic deformation and phase transformation processes of materials under shock compression, in this paper we focus on the detailed structures of shock front, the influencing factors on the shock-front structure and the atomistic processes of plastic deformations or phase transformations occuring in the shock-front regions.First, we investigate the method to construct metallic nc samples for MD simulations. There are two procedures in this method. First, we construct the initial nc samples by Voronoi geometrical method. Secondly, the initial samples are relaxed to reduce the high energy and stress caused by the unphysical grain boundaries (GBs). The residual internal stress is employed to define the difference between the simulated and the experimental synthesized samples for the first time. The initial nc samples generated by Voronoi method can reach their local minimum energy states after quenching about 10ps, and there is no evidence that much longer quenching time can reduce the energy and stress further. Then the simulated annealing at ambient temperature and stress for 40 ~ 100ps is needed to approach the lowest global energy and the lowest residual stress state. It is not necessary to anneal the samples longer. We find that the higher annealing temperature is set, the less annealing time is needed, but the annealing temperature must not be higher than 65% of the melting point. Meanwhile, we monitor the structures of GBs, the temperature descending processes and the local distributions of averaged internal stresses and the energies of the samples during the relaxation processes, and calculate the elastic constants of the nc samples under various conditions at the end of relaxations. The results show that the energy, residual internal stress and elastic constants of the computer generated samples at final state are close to the experimental preparation of nc metals.Secondly, the effects of GBs on the shock-front structures and the mechanisms of plastic deformations of nc aluminum under shock compression are qualitatively investigated, and the quantitative relationship between shock stress and width of shock front of nc copper is also studied. The influences of GBs-related plastic deformation on the shock-front structures are obtained by observing the shock stress and particle velocity profile and the atomistic views of plastic deformation processes. The quantitative relationship between the shock-front width of nc copper and the shock stress can be calculated by a series of MD simulations at different shock compression conditions, and it is almost alike to other MD simulation results. The results show that: after the elastic wave generates, the GBs sliding and deformation dominated the early plastic deformation procedure, then the partial dislocations are nucleated at the deformed GBs and spread within grains, finally the process of stacking faults, deformation twins and full dislocations forming in grains dominated the latter part of the plastic deformation. The structural characteristics after the shock front swept over are that the stacking faults and the deformation twins are left in grains and the majority of the full dislocations are annihilated at the opposite GBs. We also find that the contribution of GB-mediated plasticity to the shock-front is comparable to dislocation-mediated plasticity for nc aluminum.Then, a comparative study of plastic deformation mechanisms and the shock-front structures between nc aluminum and copper with the same texture and grain size is carried out. By comparing the one-dimensional (1D), 2D shock wave profiles and the atomistic views of plastic deformation processes of aluminum and copper at the same shock stress and strain, we find that the rise rate of nc copper in the elastic deformation is higher than that of nc aluminum. The reason may be that the lattice constant of aluminum is slightly larger than copper, leading to a more extended deformation range before softening, it will need more time to compress aluminum atom than copper. We also find that the duration of GB-dominated plastic deformation of nc copper is shorter than that of nc aluminum, and that the contribution to shock-front width from the GB-dominated plastic deformation is smaller than that of nc aluminum. The overall width of the shock front of nc copper is less than the width of nc aluminum, and the dislocation densities of nc copper are higher than the densities of nc aluminum. The mechanisms of plastic deformation for nc copper and aluminum are also different slightly, that is to say, we have observed partial dislocations, full dislocations and deformation twins in nc aluminum, but only partial dislocations are observed in nc copper. On the other hand, we find some GB-related plastic mechanisms like GB sliding, GB thickening and GB bending in nc aluminum, but no GB bending is found in nc copper. The main reason for these differences of plastic mechanisms between these two nc ductile metals is that the intrinsic parameters, such as lattice constants, stacking fault energies and shear modulus and so on result in a higher critical resolved shear stress for the leading dislocation emission of aluminum than that of copper, and finally lead to less dislocations emitted and more GB-related mechanisms and wider shock-front widths in nc aluminum.Finally, the shock-induced phase transformations of nc irons with different grain sizes are investigated. The study shows that the critical shock stress for shock-induced phase transformation of nc irons is about 15 GPa. Under shock compression, the nc irons first experience elastic deformation, then plastic deformation purely caused by grain boundaries, after that phase transformation nucleated mostly at the grain boundaries, and finally nucleation areas expanding into the entire samples. These processes can be reflected by the stress and particle velocity profiles, and also be distinguished by local atomic structures analyses in the corresponding areas. Most of the bcc irons atoms transform to hexagonal-closed packed (hcp) structure except the GB atoms. The major microstructures of the nc iron samples at final state are GBs and hcp atoms with small amount of fcc atoms as the twin boundaries. There are more GB atoms for the larger grains in the range of grain sizes studied in this paper, and more bcc iron atoms transform to hcp iron atoms, and more energy are dissipated during phase change, and turn out to bring down the final shock stress.This study primarily discuss the effects of GB which is a type of prevalent defect in materials on the mechanisms of plastic deformation and phase transformation of ductile metals, and how to further affect the typical macroscopic response of materials subject to shock compression–– shock front. The results can improve the understanding of physical procedure of the plasticity and phase transformation, and also have reference significance for revealing the links between macro-micro processes at nanosecond- micrometer scale and building the physical images of micro-processes. Although some quantitative results presented in this study for nc metals are different to the results in shock compression experiments for traditional coarse-grained (cg) metals, but the understanding achieved in this paper is also helpful to investigate the dynamic responses of ordinary cg metals, such as interpretations of microscopic mechanisms of the shock-front structures.更多还原