Multi-scale modeling of complex materials
The research activities of the Sierka group focus on the development and applications of multi-scale computational methods for investigating structure, properties and reactivity of complex materials – nanoparticles, thin films, surfaces and interfaces. Many chemical and physical properties of these materials arise from processes and features at multiple scales, both spatial and temporal. Therefore, our work involves simulations of material properties using information or models from different levels of theory: quantum mechanics, molecular mechanics and dynamics, mesoscale and continuum mechanics levels.
The spectrum of the methods currently developed in the group ranges from quantum chemical methods for extended systems over combined quantum mechanics – molecular mechanics (QM/MM) and quantum mechanics – quantum mechanics (QM/QM) approaches, to global structure optimization algorithms. These methodological developments are applied within research projects conducted in close collaboration with experimental groups from different disciplines of chemistry and physics.
Simulation methods for large molecules, surfaces and solids
The basis for our research projects within this area is the TURBOMOLE quantum chemical program package, initially developed in the group of Reinhart Ahlrichs at the University of Karlsruhe and at the Forschungszentrum Karlsruhe. With almost 20 years of continuous development TURBOMOLE has become a valuable tool used by academic and industrial researchers. It is used in research areas ranging form materials science, inorganic and organic chemistry to various types of spectroscopy, and biochemistry.
Our research in this area is devoted to the extension of the methods available within the TURBOMOLE program to periodic systems such as surfaces, interfaces and bulk solids. The main features of this new implementation are sparse storage of real space integrals and density matrices, the use of resolution of identity (RI) approximation and hierarchical approaches for numerical integration of exchange-correlation terms within density functional theory methods. The key component is the new formulation of RI approximation for the Coulomb term, which treats molecular and periodic systems of any dimensionality on an equal footing. This project plays a crucial role in future developments and applications of the TURBOMOLE program package to surfaces, interfaces and bulk systems.
Our recent publications within this project are:
- resolution of identity method for molecular and periodic systems: J. Chem. Phys. 2009, 131, 214101-1-214101-6.
- Linear scaling hierarchical integration scheme for the exchange-correlation term in molecular and periodic systems: J. Chem. Theory Comput. 2011, 7, 3097–3104.
Global structure optimization methods
Our research within this area is devoted to the development of global optimization methods and their application for design of novel materials. In general, efficient structure optimization methods are important prerequisite for computational studies of structure and properties of materials. Local optimization methods locate the nearest local minimum or a saddle point and need a reasonable initial starting point. Global optimization methods are able to locate the global energy minimum independent of the initial structure. Therefore, such methods are well suited for the design of novel materials and for structure determination of systems, which are difficult to access experimentally. The DoDo program package developed within this project uses genetic algorithm (GA) as the global optimization method. It proved efficient for automatic structure resolution of both molecular systems, surfaces and interfaces. The current application area within this project is the design and structure determination of novel low-dimensional materials by a combination of calculations and experiments.
Our recent publications within this project are:
- Review article: Prog. Surf. Sci. 2010, 85, 398-434.
Hybrid methods for materials simulations
Many chemical and physical properties of materials arise from processes and features at multiple scales, both spatial and temporal. Usually, for computer simulations at each scale a particular level of approximations is used. The usual classification of the levels is:
- quantum mechanics for the electronic scale,
- interatomic potential functions and molecular dynamics for the atomic scale,
- mesoscale or nano level for the scale of molecules and group of atoms,
- continuum models,
- finite element and finite difference methods.
The hybrid methods developed within this project allow for simulations of material properties or behavior using methods from different levels simultaneously. As an example, we have created a combined QM/MM approach that describes the reactive part of a chemical system quantum mechanically (QM) and its environment by analytical potential functions or force fields (MM). The QMPOT program which implements this method allows for a flexible combination of different software packages and includes an efficient optimizer for both and for transition structures. This way it is possible to locate transition states in systems with thousands of degrees of freedom – an essential prerequisite for the study of chemical reactions and processes in extended systems. Current development of QMPOT involves cooperation with the Scienomics Company.
Our recent publications within this project are:
- Overview of applications of the QMPOT program: Handbook of Materials Modeling, Vol. 1; S. Yip (Ed.); Springer, Dordrecht, 2005, 241-258.
- Electronic embedding: J. Chem. Phys. 2009, 130, 174710-1-174710-11.
