Computational Electronic Structure Theory (CEST)
Please visit http://old.physics.aalto.fi/en/groups/cest/ for more information.
The electronic structure gives us an atomistic view on matter that is important for many applications. Examples are materials for clean energy production, light-emitting devices (LEDs) or information and communication technologies (ICT). Perturbing the electronic structure, as done in spectroscopy, reveals more information about matter. We develop and use theoretical spectroscopy methods to probe the properties of molecules, molecules on surfaces, nanostructures, as well as semiconductors and their surfaces. We also investigate data as new resource in materials science. We participate in the development of a large scale materials database and study the potential of database driven materials science.
For more information, see the Publications list. To learn more about the CEST group, see Members.
Theory and method development
The CEST group is developing computational electronic structure theory and machine learning methods to solve pertinent problems in physics, chemistry and materials science. Our method development facilitates a better description of matter and unlocks new applications. However, new systems pose new challenges, which again require new method development.
We mainly work with quantum mechanical first principles. For the atomic and electronic structure we use density-functional theory (DFT) and for excited states (theoretical spectroscopy) Green's function methods, such as the GW approach. On the DFT side, we actively develop hybrid density-functionals that mitigate self-interaction errors, while staying computationally tractable. On the Green's function side, we are investigating approaches that go beyond GW such as second-order screened exchange self-energies. We also seek the connection between DFT and many-body Green's functions theory to devise advanced density-functionals, for example functionals that lead us beyond the random-phase approximation (RPA).
Some of our current projects are:
Core level spectroscopy is a powerful tool to characterize molecules, liquids and adsorption processes at surfaces. Accurate computational methods to predict core excitations are important for the interpretation of experimental results. However, the reliable computation of core levels remains a challenge. We are developing new approaches to predict core level spectra for molecular systems.
Strongly-correlated electrons are those which cannot be described by independent particle methods like DFT. Besides their importance to fundamental physics, strongly-correlated materials are becoming increasingly relevant for technological applications. We are researching new approaches to strong correlation that combine the best features of many different theories.
For wide dissemination, we implement most of our developments into the Fritz Haber Institute ab initio molecular simulations package (FHI-aims). If you are interested in using FHI-aims for your own work or if you would like to contribute to FHI-aims, please contact us.
More on machine learning methods is given on the Machine Learning page.
Piecewise linearity in the GW approximation for accurate quasiparticle energy predictions, M. Dauth, F. Caruso, S. Kümmel, and P. Rinke, Phys. Rev. B 93, 121115(R) (2016)
Random-phase approximation and its applications in computational chemistry and materials science, X. Ren, P. Rinke, C. Joas, and M. Scheffler, J. Mat. Sci 47, 7447 (2012)
For more information please contact Patrick Rinke.
Hybrid perovskite photovoltaics
Hybrid perovskites are a novel materials class that has surprised the photovoltaic community with its rapid efficiency increase. Since their discovery in 2012, the light-to-energy conversion efficiency of hybrid perovskite photovoltaic cells (HPPVs) doubled and rose to 22% mark - approaching conventional inorganic-based single-junction cells such as CdTe, CIGS, and polycrystalline Si. HPPVs are promising for emergent solar-cell technologies not only because of their high efficiencies, but also their low production cost and structural flexibility.
Our group is modelling HPs (CH3NH3PbI3 and other isostructural halide perovskites) and HP-based model systems to understand the materials properties and fundamental processes. The ultimate aim is to support the design of novel hybrid photoactive materials. We use density-functional theory (DFT) and DFT-based atomistic thermodynamics to investigate the atomic structure and motion (including phase transition), electronic structure, charge-transport processes as well as surface and interface physics and chemistry. We are also establishing a multiscale approach to model the structural complexity of hybrid perovskites (as a result of the irregular alignment of CH3NH3 dipoles at finite temperatures) and its impact on the materials properties.
Related summer projects:
Pair modes of organic cations in hybrid perovskites: Insight from first-principles calculations of supercell models, J. Li, J. Järvi and P. Rinke, arXiv:1703.10464.
Atomic structure of metal-halide perovskites from first principles: The chicken-and-egg paradox of the organic-inorganic interaction, J. Li and P. Rinke, Phys. Rev. B 94, 045201 (2016).
For more information please contact Jingrui Li.
Data driven materials science
For materials scientists, recent technological developments in big data and AI provide unprecedented opportunities for materials discovery and design. Nevertheless, the field of big data driven materials science is still in its infancy, with few players and limited applications. Taking advantage of big data in this area will require a paradigm shift in the institutions, culture and practices of science.
To address how novelty in science emerges, and what dynamic processes spur its development from margins to mainstream, this project conducts a unique,interdisciplinary and real-time inquiry into the development of this new trend in science from margins to mainstream, and its effects on the practices, culture and identity of material scientists.
If you are interested in the implications and/or applications of data driven materials science in education, research or industry, please contact us.
People involved: Amber Geurts
Novel Materials Discovery Laboratory (NOMAD)
Our group is also involved in the Novel Materials Discovery (NOMAD) Laboratory. In NOMAD we are developing a materials encyclopedia and data-analytics tools for materials exploration and discovery. Currently the NOMAD database already contains over 40 million total energy calculations and we are actively developing parsers to upload data from a variety of different electronic structure codes.
People involved: Lauri Himanen
Research into future technologies has come to focus on miniature multi-material devices, with the intent of harnessing quantum mechanical phenomena to perform an automated function such as generating signals or separating atoms or charges. Organic and inorganic materials are frequently employed side-by-side to take advantage of their unique capabilities. While the properties of these bulk materials are known, it is not always possible to predict or measure what occurs at the boundary between them inside a device.
Graphene Interface Nanostructures and Properties
Graphene is an immensely versatile material, but future applications will depend on how well it will interface with molecular or crystalline materials. In the study of F4TCNQ/graphene/Ir(111), we are researching the many-body physics of strongly-interacting molecular states coupled to a continuum of weakly-interacting states in graphene.
People: M. Dvorak, collaboration with P. Liljeroth (Aalto, Finland).
Organic/inorganic thin film interface optimisation
We are applying artificial intelligence (AI) methods for complex optimization to learn the interface structure of C60 molecules on the (101) surface of TiO2 anatase. These materials are frequently used in optoelectronic devices; by optimising the interface structure and properties, we aim to improve the efficiency of organic-based solar cells.
Oligothiophenes at nano-structured surfaces
Oligothiophenes are promising materials for optoelectronic devices. To increase device performance, we must understand how molecular packing structure in devices affects the electronic properties of the material. We investigate the packing structure and the electronic properties of α-sexithiophene on the nanostructured Au(100) surface by combining DFT calculations with experimental STM measurements.
People: L. Scarbath-Evers, M. Todorović, collaboration with D. Sebastiani (Halle, Germany).
Our group is part of the Collaborative Research Centre (SFB 951) Hybrid Inorganic/Organic Systems for Opto-Electronics (HIOS) of the German Science Foundation.
Integer versus fractional charge transfer at metal(/insulator)/organic interfaces: Cu(/NaCl)/TCNE, O. T. Hofmann, P. Rinke, M. Scheffler and G. Heimel, ACS Nano 9, 5391 (2015).
Space Charge Transfer in Hybrid Inorganic-Organic Systems, Y. Xu, O. T. Hofmann, R. Schlesinger, S. Winkler, J. Frisch, J. Niederhausen, A. Vollmer, S. Blumstengel, F. Henneberger, N. Koch, P. Rinke, and M. Scheffler, Phys. Rev. Lett. 111, 226802 (2013)
For more information please contact Milica Todorović.
Machine Learning in Materials Science
Machine learning is a branch of artificial intelligence and can be used as an efficient approach to predict many different properties of materials and surface stuctures. The idea is to develop a machine learning model that learns effectively from information already available for thousands of other materials by creating statistically optimized relationships between the given data. Once the model is trained sufficiently, it can make predictions for new materials with almost the same accuracy as conventional computational methods, but in only a fraction of the time and with a fraction of the computational effort. Machine learning is therefore a useful tool to speed up the quest for improved and novel materials.
Bayesian Optimisation Structure Search (BOSS)
We coupled total energy atomistic simulation methods with the Bayesian Optimization artificial intelligence technique to learn complex energy landscapes, stable structures and favourable properties. This has allowed us to understand large-scale interface configurations between organic molecules and crystalline substrates used in thin-film electronic devices.
Machine Learning Methods for Spectra of Novel Materials
The aim of this project is to develop machine learning models based on kernel ridge regression and deep neural networks that utilize the abundance of already available theoretical and experimental spectroscopic data. The models will be able to make instant predictions of spectra at negligible computational cost, thereby greatly accelerating the spectroscopic analysis of chemical structures and the discovery of entirely new materials.
The Novel Materials Discovery Laboratory
Our group is also involved in building the Novel Materials Discovery (NOMAD) Laboratory, which provides homogenized data for machine learning applications and and a notebook environment and software tools for running machine learning tasks on this data.
People involved: Lauri Himanen