Phase Field Simulation Of Microstructure Evolution During The Diffusion Process Of Carbon,nitrogen And Oxygen In Metals And Ceramics

The evolution of the microstructure of materials can usually be considered to be controlled by element diffusion.The transport of carbon,nitrogen,and oxygen are relatively common examples.Therefore,it is necessary to study the transport process.Among the research objects that can simultaneously include the diffusion and transport processes of carbon,nitrogen,and oxygen,heat treatment is a more typical type.In the current simulation methods for heat treatment,most of them are difficult to track the evolution of sharp interfaces.The phase field method can avoid these defects by simulating the diffusion interface changes.Therefore,phase field simulation is a suitable method for studying this subject.This paper establishes a generalized phase field model for the diffusion of carbon,nitrogen and oxygen in metals and ceramics.The diffusion of oxygen in ceramics and the diffusion of carbon and nitrogen in metals are compiled as software.The simulation software for the evolution of the precipitates in the process has realized the accurate simulation of the shape,size and distribution of the precipitates in the above process,and the simulation results are consistent with the previous results.A phase field model was constructed for the exsolution behavior of doped elements caused by the transport of oxygen vacancies in ceramic oxides.At the same time,the thermodynamic free energy functional is established based on the ternary regular solution model,and the numerical simulation is realized by the semi-implicit Fourier spectrum method.The results show that as the evolution time increases,the interaction energy coefficient increases and the initial concentration of doping elements increases,the number,size,surface coverage,and total length of particles all increase significantly.In a low oxygen partial pressure environment,the nanoparticles are exsolved on the surface of the A-position vacancy perovskite matrix,and then re-dissolved in a high oxygen partial pressure environment.The evolution law obtained by simulating the precipitation behavior and re-dissolution process is consistent with the experimental observation results.Aiming at the behavior of spinodal decomposition of 38 Cr Mo Al surface layer to produce low-nitrogen phase nanoparticles,a phase field model was established.Through the comparison of electronegativity,the influence of alloying elements is transformed into the influence of Cr element,so that the material system is regarded as the Fe-Cr-Al ternary system in the calculation.According to the pseudo-binary model,the free energy functional is established,and the simulation is carried out separately for whether it is a gradient concentration field and whether the elastic strain energy is considered.The simulation results show that the precipitation behavior of the low-nitrogen phase conforms to the law of spinodal decomposition.When the concentration field is changed from a uniform field to a gradient field,the incubation period of the precipitated phase will be shortened.The elastic strain energy considered in the model will strongly delay the precipitation of particles,and the precipitated low-nitrogen phase nanoparticles show obvious directionality and have a smaller particle size.The morphology of the precipitates produced by the simulation is similar to the experimental observation.A phase field model was established for the behavior of 20cr2Ni4 A that precipitates network carbides during carburizing.According to the carbon concentration distribution in the carburizing process,combined with the phase field simulation,the evolution law of the network carbide precipitation during the carburizing process is obtained.The results show that the carbide precipitation phase presents a gradient precipitation from the surface to the inside,which is consistent with the distribution of the carbon concentration gradient;when the carbon potential is lower than 0.85 wt.%,the network carbide precipitated along the grain boundary will be suppressed;in addition,increasing the system's interface energy(gradient energy coefficient greater than 1)can also inhibit the precipitation of network carbides along the grain boundaries.Based on the precipitation phase field simulation data of carbon,nitrogen and oxygen diffusion phases,a neural network model that can describe the evolution of precipitation phases is trained,and then input a small amount of distance correlation function data to gradually predict the distance correlation function value of the unknown structure.The results show that inputting two sets of known interstitial atom diffusion distance correlation function data can predict 140 steps of unknown distance correlation function values,and the prediction accuracy of the machine learning model reaches95.78%.