研究目的
To demonstrate a rigid scissors-like GW self-consistency approach (ˉ?GW0) that can be implemented at zero additional cost for large scale one-shot G0W 0 calculations, improving accuracy for large systems.
研究成果
The ˉ?GW0 method provides a computationally efficient and accurate approach for improving quasiparticle energies in large systems, with negligible additional cost beyond one-shot G0W 0 calculations. It shows promising agreement with high-level methods and experiments for large molecules and periodic systems, though its accuracy diminishes for smaller systems and electron affinities.
研究不足
The ˉ?GW0 approach, while accurate for large systems, shows deviations for small molecules due to the strong dependence of the self-energy diagonal elements on state energy, violating the rigid shift assumption. Additionally, the method's accuracy for electron affinities (LUMO) is not as good as for ionization potentials.
1:Experimental Design and Method Selection:
The study employs a rigid scissors-like GW self-consistency approach (ˉ?GW0) for large scale one-shot G0W 0 calculations, focusing on improving quasiparticle energies for large systems.
2:Sample Selection and Data Sources:
The method is tested on molecules (e.g., nitrogen, ethylene, urea, naphthalene, tetracene, hexacene) and periodic semiconductors and insulators with large supercells (6192 valence electrons).
3:List of Experimental Equipment and Materials:
Computational methods include stochastic G0W 0 and deterministic self-consistent GW approaches, with comparisons to CCSD(T) and experimental values.
4:Experimental Procedures and Operational Workflow:
The ˉ?GW0 method is applied as a post-processing step to G0W 0 calculations, requiring only two self-energy matrix elements for the HOMO and LUMO states, enabling a self-consistency cycle with negligible additional cost.
5:Data Analysis Methods:
The accuracy of ˉ?GW0 is assessed by comparing ionization potentials and fundamental bandgaps with high-level methods (CCSD(T), evGW 0) and experimental values.
独家科研数据包,助您复现前沿成果,加速创新突破
获取完整内容