Theoretical Study On The Mechanical And Thermal Properties On Several Classes Of Light Element And Light Element Compounds

New materials are essential for the development of high-end manufacturing industry and can be considered as the main carrier of high-tech development.The design of new materials will provide the means to impact many technologies and applications.Recently,first-principles calculations based on density functional theory have become a basis for material design,and their related theories and methods are well accepted.Multifunctional materials with excellent mechanical,thermal and thermodynamic properties have attracted great interest,and shown good application prospects in many fields such as machining,aerospace,precision instruments.The key to design such functional materials theoretically lies in the deep understanding of its lattice dynamic behavior,which directly determines the core technical indicators such as the stability,elastic constant,thermal conductivity and so on.Therefore,based on the lattice dynamics theory,this thesis systematically reports the mechanical,thermal and thermodynamic properties of several unreported materials composed of light elements or light elements compounds by density functional theory combined with high-throughput computing and machine learning methods.This thesis also discusses their potential applications in super-hard materials,thermoelectric materials,and thermal management materials.The specific findings are as follows:(1)We designed two types of metallic carbon phases,namely O-type and T-type carbon allotropes,from bottom-up self-assembling using diamond nanostripes as building block and C=C bond as linkers.These allotropes are energetically more favorable than most previously identified three-dimensional metallic carbon allotropies.The bulk modulus increases with the increasing of sp3 carbon in both O-type and T-type carbon allotropes,while the metal ductile will increases with the decreasing of sp3 carbon.Electron structure calculations also illustrated their anisotropic metallic conductivity,high Fermi velocity,and tunable electronic properties by mechanical strain.(2)Based on the first-principles method and Debye-Einstein model,the lattice configuration,stability,phonon spectrum,elastic modulus,thermal-elastic modulus and thermal expansion properties of layered ternary transition metal borides(MAB)were studied by systematically high-throughput calculations.Among these materials,we found Ti2Al2B2,SC2AlB2,Ti2AlB2,Zr2AlB2 and Hf2AlB2 are energetically more favorable than those of synthesized MAB phases.SC2AIB2,W2AIB2 and TC2AIB2 have better elastic modulus properties,and Mln2Al2B2,Tc2Al2B2,Nb2AlB2,W2AlB2,TC2AlB2,CO2AIB2 and Ni2AlB2 are more ductile than that of synthesized MAB phase.In addition,we found that these layered MAB phases have low thermal expansion as refractory solid in 300?1500 K.The desired bulk modulus decreases by only 10%in 1500 K,which demonstrated the good thermal shock resistance.(3)Inspired by the similarity of the structure and bonding between ice phase and silicon phase,we proposed four self-assembling low-density silicon clathrates,named Si-CL-A,SiCL-B,Si-CL-C and Si-CL-D,based on the same bonding topologies of clathrate hydrates.In the low-density region of phase diagram,Si-CL-B,Si-CL-D,and Si-CL-C would overtake diamond silicon and emerges as the most stable Si allotropes.Further electron band calculation showed that they have suitable direct band gaps(?1.1 eV)and small effective masses,which are comparable to the GaN semiconductor.Using Slack model,we further calculated the thermal conductivity of these four silicon phases,which are two orders of magnitude lower than that of diamond silicon.The ultralow mass density,and strong anharmonic phonon interaction are responsible for the low thermal conductivity.(4)We applied a machine learning method for thermal conductivity prediction.First of all,we constructed the thermal conductivity data set of 231 existing materials,then we constructed and trained an artificial neural network(ANN).During the training process,8 basic lattice dynamic properties of materials,such as V,M,n,np,B,G,B' and G',are used as descriptors(inputs)and the measured thermal conductivity are used as targets(output);then the mapping model between descriptors and thermal conductivity are construct.The trained ANN model shows the capability of accurately predicting thermal conductivity spanning 4 orders of magnitudes,and shows the ability of extrapolation prediction.Compared with the widely used Slack model,our neural network predicting model has higher reliability.In addition,the heat transport properties of other 3892 materials are further discussed using the trained neural network model.(5)The experimental findings of two-dimensional(2D)magnetic semiconductors provide an ideal platform to study the spin phonon coupling(SPC)effect of low dimensional systems.By first-principles calculations,we conducted a comparative study of lattice structures,phonon dispersions,lattice thermal conductivities,heat capacities,group velocities,scattering rates,phonon anharmonicity,phonon magnon scattering and magnetic coupling parameters for 2D magnetic transition metal trihalides semiconductors MX3(M=Fe,Ru;X=Cl,Br,I)with and without introducing spin.The results shown that SPC can affect the thermal transport properties of two-dimensional magnetic semiconductors,which induce anomalous lattice thermal conductivity instead of following typical 1/T relation and show significant composition effect.For Fe tri-halide monolayers,a significantly lattice constant increasing and phonon mode softening are observed.Comparing to Fe tri-halide monolayers,the SPC in Ru tri-halides is relatively weaker.The investigations of magnetic coupling parameters show deep response with lattice stretching/compressing,which imply that the important role of SPC in stabilizing the ferromagnetic ordering in all studied 2D magnetic semiconductors.