As a key material for solar photovoltaic power generation, the purity of silicon plays a crucial role in the photovoltaic conversion efficiency and performance. The electron beam melting (EBM), which is irreplaceable metallurgical methods, has been widely used for silicon purification. However, high energy consumption result in relatively high cost, which limits its widespread application in melting industry. This study was centered on the energy distribution that proposes a new approach to reducing energy consumption during EBM. The scanning form of electron beam is most important factor concerning energy consumption. Different scanning forms can affect the characteristics of molten pool which is closely correlated with the purification process. In this study, efficient and energy-saving preparation processes of EBM are being explored through the numerical model and experimental. The relationship of energy distribution form and purification efficiency is studied. The research results indicate that electron beam scanning under uniform distribution can achieve more uniform temperature and flow field distribution in the molten pool, which has a higher energy utilization rate during silicon purification by EBM. The mathematical model established in this study can effectively predict the energy consumption level under different melting parameters, providing important reference for the optimization and energy conservation of EBM process for other materials. In addition, evaporation characteristics of volatile impurity, migration of the second phase particles in the melt, controlling of melt pool characteristics and the crystal growth control all can be conducted based on this study.