Delocalization Error in DFT-Predicted Extreme Long-Range Functionalization of Carbon-Doped Hexagonal Boron Nitride
Autor: | Benjamin G. Janesko |
---|---|
Rok vydání: | 2019 |
Předmět: |
Materials science
Hydrogen Dopant Doping chemistry.chemical_element 02 engineering and technology 010402 general chemistry 021001 nanoscience & nanotechnology 01 natural sciences Diatomic molecule Molecular physics 0104 chemical sciences Surfaces Coatings and Films Electronic Optical and Magnetic Materials Condensed Matter::Materials Science Delocalized electron General Energy chemistry Physics::Atomic and Molecular Clusters Molecule Density functional theory Physics::Chemical Physics Physical and Theoretical Chemistry 0210 nano-technology Carbon |
Zdroj: | The Journal of Physical Chemistry C. 123:15062-15070 |
ISSN: | 1932-7455 1932-7447 |
DOI: | 10.1021/acs.jpcc.9b03159 |
Popis: | Delocalization error is a serious failure of approximate density functional theory (DFT). This work shows how delocalization error affects the predicted catalytic activity of carbon-doped hexagonal boron nitride. While the charge and spin of an isolated carbon dopant are themselves predicted to be localized, nonhybrid DFT approximations predict spurious long-range transfer of charge and spin from such dopants to distant hydrogen adatom and molecular oxygen adsorbates. This effect mirrors nonhybrid DFT approximations’ predicted spurious long-range charge transfer in other systems, for example, dissociated heteronuclear diatomics Liᵟ⁺–Fᵟ⁻. The effect is robust to simulation details, occurring in finite-cluster, one-dimensionally periodic, and plane-wave periodic supercell simulations. In extreme cases, a single carbon dopant is predicted to activate molecular oxygen adsorbed 50 nm away and activate an isolated O₂ molecule suspended 1 μm away. The effect is traced to nonhybrid approximations’ delocalization error, which produces incorrect convex curves of energy as a function of fractional charge transfer from the carbon dopant to molecular oxygen. Long-range-corrected hybrid DFT approximations mitigate these delocalization errors and predict that carbon dopants will only activate adjacent molecular oxygen. This motivates adoption of long-range-corrected hybrids in future DFT simulations of doped boron-nitride nanostructures. |
Databáze: | OpenAIRE |
Externí odkaz: |