Investigation Of Several Energy Materials Based On First-principles Calculations And Machine-learning Potential

With China’s 2060 carbon neutrality target,in order to reduce the demand and dependence on nonrenewable fuel source,and decrease carbon emissions,exploring new energy storage and conversion materials is currently the top priority in the research filed.Based on the urgent needs for the development of energy materials,this thesis combines firstprinciples calculation and machine-learning potential functions method to investigate several representative energy materials,and will provide a certain theoretical basis for the energy application.The main research contents of the thesis include:(1)Research on the thermoelectric properties of two-dimensional SiTe materials.Design QL-SiTe with quadruple layer,α-SiTe with blackphosphorene-like structure and β-SiTe with blue-phosphorene-like structure,and study the electronic,phonon,and thermoelectric transport properties of 2D SiTe with three different structural phases.Among these three structure phases,β-SiTe possesses the highest Seebeck coefficient and further a high power factor.Strong phonon scattering heavily influences the lattice thermal conductivity of QL-SiTe and β-SiTe.Therefore,the ZTmax value of β-SiTe can reach 0.95 at T=1300 K for both n and p-type doped systems,suggesting good potential for hightemperature heat conversion.(2)Research on strain engineering the electronic and phonon transport properties of two dimensional InSe.The first-principles calculation results reveal the excellent charge transport ability of InSe monolayer with the carrier mobility exceeding 1000 cm2V-1s-1.Nevertheless,the tensile strain has a limited effect on the effective mass and carrier mobility.In the case of phonon transport properties,the anharmonic phonon scattering can be enhanced with the increasing tensile strain,giving rise to the higher phonon scattering rate,lower phonon group velocity and heat capacity,and therefore thermal conductivity decrease from 25.9 to 13.1 W/mK in the 6%strained system.(3)Research on the quasiparticle electronic and optical properties of two-dimensional group-Ⅳ tellurides.All group-Ⅳ tellurides are indirect bandgap semiconductors except for monolayer PbTe,and all of them are predicted to have low carrier effective masses.The study of the excitonic effect shows that the electron-hole interaction has a significant impact on the optical properties of the monolayer group-Ⅳ tellurides,and the highest exciton binding energy of 0.598 eV is predicted for SiTe.Interestingly,the physical properties of monolayer group-Ⅳ tellurides are subject to an isotropic trend:from SiTe to PbTe,the differences of the calculated quasiparticle band gap,optical gap and further exciton binding energy along different directions tend to decrease.The atomic bonding analysis reveals that more electrons are localized around the Te atoms with the increase of atomic number,instead of between the Te and group-Ⅳ atoms,the more ionic the bonding becomes.However,no significant Coulomb repulsion from non-bonding electron pairs is observed,and further the structure shows no anisotropy,leading to isotropic electronic properties.(4)Research on the simulation of LLZO solid electrolyte based on machine-learning models.We developed a machine-learning model that enables the large-scale atomistic simulation of LLZO structures with near first-principles accuracy.This model is based on a first-principles reference dataset that includes the calculated energies of LLZO,Al-doped LLZO and Ga-doped LLZO by using the artificial neural network(ANN)method.By comparing with DFT calculations,the obtained RMSE values of the test set for Al-LLZO and Ga-LLZO potential are 6.42 meV/atom and 9.08 meV/atom,respectively.The optimized potentials are used for the molecular dynamic simulations for the complex LLZO system,and the calculated Li ion conductivities have an excellent agreement with the measured values in experiments.The main innovations of the thesis include:(1)Two dimensional β-SiTe is predicted to be a promising thermoelectric material,with a balanced energy conversion performance for both n and p-type doped systems,suggesting the great potential on the application of thermoelectric generators.(2)With the calculated quasiparticle electronic and optical properties,we discovered the isotropic trend of the crystal structure and the physical properties of monolayer group-Ⅳ tellurides,and reveal the underlying mechanism of the more isotropic behavior.(3)We constructed a machine-learning potential function with near first-principles accuracy for the LLZO solid electrolyte based on the ANN method.This potential can be employed for molecular dynamic simulations of the large-scale system,and the calculated Li ion conductivities are close to the experimental results.