<div>
<div>
<div>
<p>Interaction energies of halide-water dimers, X<sup>-</sup>(H<sub>2</sub>O), and trimers, X<sup>-</sup>(H<sub>2</sub>O)<sub>2</sub>, with X = F, Cl, Br, and I,
are investigated using various many-body models and exchange-correlation functionals selected across the hierarchy of density functional theory (DFT) approximations.
Analysis of the results obtained with the many-body models demonstrates the need to
capture important short-range interactions in the regime of large inter-molecular orbital
overlap, such as charge transfer and charge penetration. Failure to reproduce these
effects can lead to large deviations relative to reference data calculated at the coupled
cluster level of theory. Decompositions of interaction energies carried out with the absolutely localized molecular orbital energy decomposition analysis (ALMO-EDA) method demonstrate that permanent and inductive electrostatic energies are accurately reproduced by all classes of XC functionals (from generalized gradient corrected (GGA) to
hybrid and range-separated functionals), while significant variance is found for charge
transfer energies predicted by different XC functionals. Since GGA and hybrid XC
functionals predict the most and least attractive charge transfer energies, respectively,
the large variance is likely due to the delocalization error. In this scenario, the hybrid
XC functionals are then expected to provide the most accurate charge transfer energies.
The sum of Pauli repulsion and dispersion energies are the most varied among the XC
functionals, but it is found that a correspondence between the interaction energy and
the ALMO EDA total frozen energy may be used to determine accurate estimates for
these contributions. </p>
</div>
</div>
</div>
Electronic hole transfer in rutile and anatase TiO2: Effect of a delocalization error in the density functional theory on the charge transfer barrier height
