Improved Kohn-Sham DFT-based schemes
|Density functional approximations cannot accurately describe interactions between non-overlapping densities. This inherent deficiency results in errors for intermolecular (e.g. pi stacked benzene dimer) as well as intramolecular (e.g. branched alkanes, isomerisation reactions) weak interactions. An efficient solution to improve the performance of density functionals for weak interactions is to add a damped attractive long-range dispersion energy correction to the GGA, hybrid GGA or meta-GGA energy. We proposed density-dependent dispersion coefficients and an overall density-dependent correction that improve the “intramolecular errors” efficiently, without altering the long-range correction. Our most recent scheme, called dDsC, is broadly applicable to every chemical systems, reactions and is highly robust. dDsC is available in GAMESS and ADF2012 (see download).
Stephan’s latest dDsC scheme combining our density-dependent dispersion coefficients and our genuine density dependent damping factor. To see how the method performs, check the extenstive benchmarking here on JCTC2011.
dDsC is implemented in ADF2012, QChem and GAMESS. Feel free to contact us for our free patch available for GAMESS (see download).
For more details on the derivation of our density-dependent dispersion coefficients, check here: JCP2011!
For more information on our density-dependent damping factor, check here JCTC2010.
A closely related topic of interest concerns the origin of the error of standard density functional approximations for treating weak intramolecular, charge transfer interactions and mixed valence complexes.
We demonstrated that the errors of standard density functionals for intramolecular interactions are due to a combination of over-repulsiveness of the functional in the short-range and the ubiquitous missing dispersion interaction (read our featured article in TCA)!
Currently, charge-transfer complexes span the field of organic electronics making them of considerable interest. Our article in JCTC2012 reveals that standard functional fail to accurately describe interaction energies not only because of the missing long-range exchange, as generally assumed, but also as a result of the neglect of weak interaction. This paper is amongst the most read JCTC paper of the year 2012.
Interaction energy of a cofacial TTF-TCNQ complex at various levels.
Our follow-up paper on molecular precursors to organic electronics (JCTC2012b) introduce a database made of 26 radical dimer complexes that represent binding energies between organic functional units. Reproducing the energy profile of the Orel26rad test set is highly challenging and rely upon the subtle interplay between the lack of dispersion and the delocalization error.
In the last few months, we took advantage of both our dDsC scheme and the Pi29/Orel26rad database to devise a related correction for the density functional tight-binding approach. In this context, we have also addressed a dramatic error in the DFTB parametrization that prevents the use of the framework for simulating the structure and energies of non-covalent interactions involving sulfur (JCTC2013).
Our contributions summarized herein will be part of a special 2014 issue of Accounts of Chemical Research.