Abstract Grain boundary mechanics plays a key role in the strengthening of polycrystals. In this study, deformation of aluminum grains is investigated to better understand the origins of strengthening mechanisms near the grain boundaries. For this reason a sample with nearly columnar structure was prepared and fully characterized by electron back scattered diffraction measurements using the two side surfaces of the sample before loading. The total strain distribution on one surface of the sample is measured during tensile loading using the digital image correlation technique. The numerical studies include implementation of crystal based finite-element model to understand the origin of grain boundary strengthening using both local phenomenological, local dislocation density based and non-local strain gradient based constitutive laws. The effects of mesh refinement, through thickness discretization, boundary condition, and orientation scatter are investigated. Comparison of experimental and simulated total strain distributions near grain boundaries suggests the existence of strain gradient hardening. For this reason, a novel non-local flux based model is developed and implemented with a consistent time integration scheme. Dislocation density based models reveal strain distributions that are in better agreement with the experimental results than the phenomenological model that was not able to capture grain boundary strengthening and experimental strain distributions. The non-local strain gradient based strain hardening law, which does not require separate definition of grain boundary elements, was successful in capturing the strengthening effect near grain boundaries.

中文翻译：

---

铝寡晶晶界附近应变硬化的研究：实验和基于晶体的有限元方法

摘要 晶界力学在多晶强化中起着关键作用。在这项研究中，研究了铝晶粒的变形，以更好地了解晶界附近强化机制的起源。为此，制备了具有近柱状结构的样品，并在加载前使用样品的两个侧表面通过电子背散射衍射测量充分表征。在拉伸加载期间使用数字图像相关技术测量样品一个表面上的总应变分布。数值研究包括实施基于晶体的有限元模型，以使用基于局部现象学、基于局部位错密度和基于非局部应变梯度的本构定律来理解晶界强化的起源。通过厚度离散化、边界条件和方向散射研究了网格细化的影响。晶界附近的实验和模拟总应变分布的比较表明存在应变梯度硬化。出于这个原因，开发了一种新颖的基于非局部通量的模型，并使用一致的时间积分方案实施。基于位错密度的模型显示的应变分布比无法捕捉晶界强化和实验应变分布的现象学模型更符合实验结果。基于非局部应变梯度的应变硬化定律，不需要单独定义晶界元素，成功地捕捉了晶界附近的强化效应。