Efficient Automatized Density-Functional Tight-Binding Parametrizations: Application to Group IV Elements.

Area of Science:

• Computational materials science
• Quantum chemistry
• Solid-state physics

Background:

• Density-functional tight-binding (DFTB) methods offer a balance between accuracy and computational cost.
• DFTB parameter sets require element-specific determination, which can be labor-intensive.
• Existing parameter sets may lack accuracy for certain applications.

Purpose of the Study:

• To develop an efficient and automated procedure for generating accurate DFTB parameter sets.
• To improve the predictive power of DFTB for material properties.
• To create reliable parameters for group IV elements and their binaries.

Main Methods:

• An automated procedure for generating unbiased training sets.
• Optimization of DFTB parameters using a pattern search method.
• Targeting the reproduction of formation energies and forces from density-functional theory (DFT).

Main Results:

• The developed approach successfully generated new parameter sets for group IV elements and their binaries.
• These new parameters demonstrate significantly improved accuracy compared to previous sets.
• Enhanced accuracy was particularly evident in the prediction of energies and forces.

Conclusions:

• The automated procedure provides a robust and efficient way to generate high-quality DFTB parameter sets.
• The improved parameters enhance the reliability of DFTB for materials simulations.
• This work facilitates more accurate and efficient computational studies in materials science.