研究晶体的熔化和结晶在实际应用上有重要意义，但在金属和合金内难以观察内部单个粒子的运动。近20年出现的胶体晶体体系为研究熔化和结晶提供了良好的平台。低分散度、光学显微镜下可见、相互作用势可调控等优点使得胶体粒子具有丰富的相行为，并能用图像处理得到单个粒子的运动信息，为研究成核提供了丰富的信息。 本文主要探究了在胶体晶体中过热晶界附近的成核变化。首先通过反复退火制备具有较长晶界和较大晶粒的胶体晶体样品，再利用汞灯加热实现对观察区域的均匀加热，使得晶界处于过热状态，成功观察到了过热晶界的熔化现象。得到高质量的录像后，采用标准的图像处理算法进行图像处理并计算了胶体晶体的基本结构量\psi_6和径向分布函数。利用两个基本结构量结合图像处理得到了整个熔化过程中液核生长随时间的变化。 实验结果显示，晶界的成核不同于一般晶体内部的成核，实验中观察到的熔化过程液核临界半径较小或不存在临界半径，且液核生长方向和晶粒晶向有关，具体分析还有待进一步实验的进行。

The investigation into the melting and crystallization of crystal is of vital importance in the application of industry. The long-standing problem laid in the research is the limitation of observing the single particle in the crystal either in metal or alloy. It is hard to obtain the bulk information in the crystal due to the observation method and the small scale of the molecular, otherwise expensive microscopy such as TEM, SEM must be needed in the experiment. Colloidal crystal arises during recent two decade years as an ideal platform to investigate the melting and crystallization of crystal. The colloidal particle is large enough to be observed in the optical microscope and the interaction of colloidal particle can be tuned by using a variety of methods. By applying the image processing technology on the colloidal crystal we can obtain the position information of the whole system, which brings convenient to the research of the phase transition in the colloidal crystal. The purpose of this article is to investigate the nucleation of the superheated grain boundary in colloidal crystal. We manage to obtain a high-quality sample with relative long grain boundary and large domain by annealing for several times. The grain boundary is locally heated by the lamp whose focal plane is the same as the microscope and the light is tuned to be uniform by using a film. We observe the nucleation process of the superheated grain boundary. The relation between the size of the liquid nucleus and time is calculated by obtaining the 6 and radial distribution function from image processing of the video. The experiment result shows that nucleation growth of the grain boundary differs from those in the bulk of crystal. The critical radius of the nucleus observed in the experiment is relatively small or not even exist and the direction of growth is depended on the direction of crystalline of the domain. Further experiment is needed to obtain the explicit relation.