Final Report Summary - BOOSTQUANTUMCHEM (Boosting the performance of Quantum Chemistry for nanocatalysts, biomolecules and graphene layers by solving the fundamental drawback of van der Waals interactions in Density Functional Theory)
The EU funded research project BoostQuantumChem was focused on solving the basic problem of dispersion interaction within the Quantum Chemical method Density Functional Theory (DFT). Quantum Chemistry holds a unique position within the Life Sciences, as it forms a bridge between abstract theories from the field of Quantum Mechanics and innovative applied research in fields such as biomedicine, macromolecules, catalysis and material design. DFT differs from standard quantum chemical methods by its connection to the experimentally observable quantity electron density. As a result, it holds a simple promise of an exact solution to the Schrödinger equation and it is without a doubt the most widely used quantum chemical method today. The extraordinary success of DFT can be attributed to the simplistic elegance of the theory combined with a low computational cost compared to wave function based methods. The computational simplicity does not only result in the ability to solve problems faster, but more importantly, allows to tackle challenges which are not accessible to wave function based methods. For instance, DFT is, at the moment, the only available quantum chemical method which can be applied to systems of extensive size such as peptides, nanotubes and graphene layers. Unfortunately, a specific problem which severely affects the reliability of DFT is the poor description of van der Waals (vdW) interactions, in particular London dispersion, which is often a determining factor for the stability of systems.
In the framework of this project, we have developed a novel BH-DFT-D method which enables to correct the performance of DFT for the description of dispersion interactions. Since the main challenge of DFT in describing dispersion interactions lies in the local character of the popular exchange-correlation functionals, the basic idea behind the BH-DFT-D method is to add an energy correction which is evaluated from non-local information. The analytical expression of the energy correction has been derived from intermolecular perturbation theory, which connects dispersion energy with atomic static polarizabilities. The strength of BH-DFT-D as a dispersion-including DFT method lies in its strong non-empirical character, as it uses only quantities derived from ab initio properties calculated on the fly. The form of the dispersion energy correction allows to retain the non-local long-range character through the use of a four-centered expression, as well as the full anisotropic character through the use of atomic polarizability tensors. Due to a developed density dependent damping function, it can be combined with any local density functional. During the project, the damping function has been optimized for the B3LYP and PBE0 functionals. For these two functionals, the BH-DFT-D method was shown to be able to reproduce interaction energies and geometries well within the boundaries of the desired accuracy of 1 kcal/mol and 0.2 Angstrom.
The BH-DFT-D method allows to perform accurate energy and geometry calculations on non-covalent bonded dimers in four straightforward steps:
1) Performing standard DFT calculations to obtain the interaction energy of the dimer and forces in the case of geometry optimization.
2) Performing an analytical DFT calculation for obtaining multipole polarizabilities of the monomers.
3) Partitioning the polarizabilities into atomic contributions using the Hirshfeld method.
4) Obtaining the BH-DFT-D dispersion energy correction and, if needed, the atomic forces.
Steps 1 and 2 can in principle be performed with any existing Quantum Chemical software and in this project we have used a locally modified version of the Turbomole program. Steps 3 and 4 are performed using programs STOCK and ATDISP developed by us, which are available as open source programs upon contact with the developers (alisa.krishtal@gmail.com). Both STOCK and ATDISP programs were optimized for computational efficiency and parallelized using the MPI protocol. The BH-DFT-D method can be applied on any system that can be straightforwardly divided into two interacting entities and for which static polarizabilities can be obtained at the DFT level.
The project was terminated prematurely after 14 months due to the resgination of the researcher for personal reasons. At the point of termination, the BH-DFT-D method reached a state where it is available for applications in a wide range of fields including catalysis, medicinal chemistry, biochemistry, physical chemistry, polymer chemistry and material science.
In the framework of this project, we have developed a novel BH-DFT-D method which enables to correct the performance of DFT for the description of dispersion interactions. Since the main challenge of DFT in describing dispersion interactions lies in the local character of the popular exchange-correlation functionals, the basic idea behind the BH-DFT-D method is to add an energy correction which is evaluated from non-local information. The analytical expression of the energy correction has been derived from intermolecular perturbation theory, which connects dispersion energy with atomic static polarizabilities. The strength of BH-DFT-D as a dispersion-including DFT method lies in its strong non-empirical character, as it uses only quantities derived from ab initio properties calculated on the fly. The form of the dispersion energy correction allows to retain the non-local long-range character through the use of a four-centered expression, as well as the full anisotropic character through the use of atomic polarizability tensors. Due to a developed density dependent damping function, it can be combined with any local density functional. During the project, the damping function has been optimized for the B3LYP and PBE0 functionals. For these two functionals, the BH-DFT-D method was shown to be able to reproduce interaction energies and geometries well within the boundaries of the desired accuracy of 1 kcal/mol and 0.2 Angstrom.
The BH-DFT-D method allows to perform accurate energy and geometry calculations on non-covalent bonded dimers in four straightforward steps:
1) Performing standard DFT calculations to obtain the interaction energy of the dimer and forces in the case of geometry optimization.
2) Performing an analytical DFT calculation for obtaining multipole polarizabilities of the monomers.
3) Partitioning the polarizabilities into atomic contributions using the Hirshfeld method.
4) Obtaining the BH-DFT-D dispersion energy correction and, if needed, the atomic forces.
Steps 1 and 2 can in principle be performed with any existing Quantum Chemical software and in this project we have used a locally modified version of the Turbomole program. Steps 3 and 4 are performed using programs STOCK and ATDISP developed by us, which are available as open source programs upon contact with the developers (alisa.krishtal@gmail.com). Both STOCK and ATDISP programs were optimized for computational efficiency and parallelized using the MPI protocol. The BH-DFT-D method can be applied on any system that can be straightforwardly divided into two interacting entities and for which static polarizabilities can be obtained at the DFT level.
The project was terminated prematurely after 14 months due to the resgination of the researcher for personal reasons. At the point of termination, the BH-DFT-D method reached a state where it is available for applications in a wide range of fields including catalysis, medicinal chemistry, biochemistry, physical chemistry, polymer chemistry and material science.