Jacob’s Ladder of DFT

By

The latest meta-GGA functional

Density functional theory is at the heart of atomistic modelling of materials. There is a huge variety of software packages that implement DFT.

The method is capable of evaluating pretty much all essential properties: mechanical, thermodynamic, electronic, optical, etc. Implementations are becoming “black box” packages, accessible to ever widening audience. BIOVIA on its own offers four different DFT-based packages!

A major question for DFT practitioners remains open:

What is the accuracy of specific DFT calculations? Can it be improved without an exorbitant computational cost?

Jacob’s Ladder

The main approximation in DFT is that of an exchange-correlation functional. We know that it is essentially a fudge, but a very successful one. And DFT can be made more accurate by including more physics in the formulation of exchange-correlation effects.

The Jacob’s ladder description refers to various levels of sophistication and hence accuracy in DFT: from local density approximation, to generalized gradient approximation (GGA), meta-GGA, hybrid functionals, etc.

The term itself was coined by Professor John Perdew – the name which is well familiar to DFT practitioners, as he is the author of many popular functionals currently in common use.

The latest addition to the collection of robust and accurate functionals is described in the paper “r2SCAN-D4: Dispersion corrected meta-generalized gradient approximation for general chemical applications” published in The Journal of Chemical Physics. This article presents a new dispersion-corrected parameterization of the meta-GGA SCAN functional – a major development that makes this functional useful for studies of weekly interacting molecular blocks. Application areas that are opening up include molecular crystals, chemical reactions in gas phase and in a solvent, molecule-surface interactions, and so on.

This new implementation is now available in TURBOMOLE, a quantum mechanics code distributed by BIOVIA.

This powerful solver has been described in detail in the last year’s review paper in The Journal of Chemical Physics, “TURBOMOLE: Modular program suite for ab initio quantum-chemical and condensed-matter simulations.”

Development and Validation

BIOVIA’s Uwe Huniar made major contributions to the r2SCAN-D4 development and validation. The new description is at least as accurate as hybrid functionals for structures of transition metal complexes, reaction energies for gas-phase reactions, and lattice energies of molecular crystals.

This latest development is available in the BIOVIA 2022 Collaborative Science Portfolio. It certainly makes TURBOMOLE ever more attractive for studies of molecular systems, including molecular crystals.

It is also likely that other solvers will adopt the new description soon, pushing the boundaries of future DFT applications.

Learn More

 

Victor.MILMAN@3ds.com'

Victor Milman

BIOVIA, Dassault Systèmes
Senior Director of the Quantum Mechanics and Nanotechnology R&D Team, Victor Milman, Ph.D., joined BIOVIA in 1994 and currently serves as a senior fellow and manager of quantum mechanics and nanotechnology research and development team. He graduated from Moscow Institute of Physics and Technology and received his doctorate in solid state physics from The Ukrainian Academy of Sciences. His subsequent research at the Institute of Metal Physics in Kiev focused on development of first principles techniques for study of lattice properties of inorganic crystals. This work continued at the Cavendish Laboratory, Cambridge, where he was employed as a Research Associate for the SERC Collaborative Computational Project in electronic structure of solids. This activity in the group of Professor Heine and Professor Payne culminated in the public release of CASTEP, a revolutionary code for quantum-mechanical modelling of solids and surfaces. Milman further worked for a year as a visiting research fellow at the DOE Oak Ridge National Laboratory, concentrating on applications of CASTEP to physics of semiconductors, from modelling growth processes to study of extended defects. Victor Milman has 150 peer-reviewed publications with the h-index of 29, which reflects both productivity and high scientific impact of his research. His contributions include numerous conference presentations, co-supervision of doctorate students with University of Cambridge and with University College London, organization of meetings and symposia, regular refereeing of papers for the major journals in physics and chemistry.
Victor.MILMAN@3ds.com'

Latest posts by Victor Milman (see all)