Pseudopotentials, an Overlooked Source and Remedy of DFT Errors.

Area of Science:

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

Background:

• First-principles calculations are crucial for materials science, but their accuracy is often limited by pseudopotential approximations.
• Existing pseudopotentials can introduce significant errors in atomic energy levels, deviating from fundamental quantum mechanical theorems.
• The impact of pseudopotentials on calculation accuracy is not well-understood or addressed.

Purpose of the Study:

• To investigate and quantify the errors introduced by common pseudopotentials in first-principles calculations.
• To develop a novel approach using atomic-level adjusted pseudopotentials for improved accuracy and efficiency.
• To benchmark the new method against established techniques using semiconductor bandgap calculations.

Main Methods:

• Development and application of atomic-level adjusted pseudopotentials.
• Investigation of the interplay between adjusted pseudopotentials and exchange-correlation functionals.
• Benchmarking bandgap calculations for 54 semiconductors containing monovalent copper (Cu).

Main Results:

• Identified significant errors in most existing pseudopotentials, affecting atomic energy levels.
• Demonstrated that atomic-level adjusted pseudopotentials offer a pragmatic correction, balancing accuracy and efficiency.
• Corrected erroneous metal predictions for 11 compounds and reduced mean relative error in bandgap calculations from 80% to 20%.

Conclusions:

• The proposed atomic-level adjusted pseudopotentials significantly enhance the accuracy of first-principles calculations.
• This approach provides a practical method to improve computational materials science predictions.
• The accuracy achieved surpasses that of standard hybrid functionals and GW methods for the studied semiconductors.