Multi-scale And Multi-dimensional Analysis Of Metallic Glass

From Stone Age,Bronze Age,Iron Age,to Information Age,our ability to understand and to change the world is,to a large extent,marked by the development of material science.In the advancing of material science,one of the most important way to have new insight or new materials is through the development of investigation methods.One example is how the development of microscope promotes our understanding of matter.In the field of metallic glasses,the development is marked by the evolution of synthesis and characterization method.Although bulk metallic glass is into the age of mass production,the most important question remains in our understanding on the nature of glassy material.Here,by applying different methods in the investigation of metallic glasses,we performed multi-scaled and multi-dimensional analysis on glass forming ability and relaxation dynamics.In this dissertation,the preparation and characterization of low-dimensional metallic glasses is discussed.By pulsed laser deposition technique,we obtained CuZrAl and CuZr films with the thickness of several atoms,and then monitored the evolution using aberration-corrected transmission electron microscope.Results showed that the number of nuclei is largely different between the two compositions,and this is due to the doping effect of Al,which explains the enhanced glass forming ability of Cu Zr Al.We also prepared PdSi nanoparticles with size range of 1-5 nm.The evolution of structure with size indicates the critical nucleus size of PdSi nanoparticle,which supports the classical nucleation theory.We also studied the microscopic origin of relaxation in glassy systems by means of molecular dynamics simulation.Interatomic potentials with different stiffness is designed and results show that these systems have different low temperature relaxation behavior.A larger stiffness correlates with a more significant low temperature relaxation.By analysis of the correlation between molar volume and bulk modulus of different alloys,we found that rare-earth based metallic glasses have larger atomic interaction stiffness,which is in agreement with our simulation results.This way we have an explanation for the generally stronger ?-relaxation of rare-earth based metallic glasses.By analysis of microscopic dynamics of these systems,we found a strong correlation between the intensity of dynamical heterogeneity and relaxation dynamics.With the fast development of machine learning technique,we applied support vector machine on the analysis of glass forming ability in binary alloys.We first built a database from preexist data,then,models with the ability to distinguish good and bad glass formers is developed using support vector machine.We also assessed the effects of different input features,this way,we found that the information of melting temperature plays the most effective role in deciding the glass formation ability of binary alloys.This dissertation,with different approaches applied,aimed to study fundamental questions of metallic glasses in multi-scaled and multi-dimensional analysis,in hope to broaden our understanding of the nature of glassy materials.