The search for organic Dirac materials


SNIC 2016/1-352


SNAC Medium

Principal Investigator:

Alexander Balatsky



Start Date:


End Date:


Primary Classification:

10304: Den kondenserade materiens fysik

Secondary Classification:

10201: Datavetenskap (= Datalogi)




The study of Dirac materials, i.e. materials where the low-energy fermionic excitations behave as massless Dirac particles has been of ongoing interest for more than two decades. Such massless Dirac fermions are characterized by a linear dispersion relation with respect to the particle momentum. In the solid-state quasi particle excitations showing a linear dispersion relation have been observed in, for example, graphene, d-wave superconductors, topological insulators or Weyl semi-metals. Our work concentrates on the identification, classification, design and suggestion of organic Dirac materials. The reasons to focus on organic materials can be motivated as follows: * In contrast to inorganic materials, there are only a few organic Dirac materials known at this time. In particular, no organic bulk topological insulators have yet been identified. *The main constituents of organic crystals are carbon, hydrogen, nitrogen and oxygen plus a low percentage of transition metal elements. Technologically relevant quantities of these substances are easily accessible. Up to now, only a few examples for Dirac materials in the class of organic crystals are present, for example, the organic conductor $\alpha$-(BEDT-TTF)$_2$I$_3$ or the organic $d$-wave superconductors $\kappa$-(BEDT-TTF)$_2$Cu(NCS)$_2$ and $\kappa$-(BEDT-TTF)$_2$Cu[N(CN)$_2$]Br . However, the potential for technologically relevant materials is unexplored. For inorganic materials high-throughput calculations and data mining have been performed to identify topological insulators from known crystal structures. Similar investigations for organic materials have not been performed yet. For the identification of organic Dirac materials we will follow a different approach based on data mining and machine learning. Thousands of experimentally studied organic crystals are present in the literature, mostly with precise structural information. By applying automatized band structure calculations based on density functional theory DFT we are building up an organic materials database. The content of the database is used for the prediction of topological insulators, Dirac and Weyl semi-metals as well as candidates for high-$T_c$ superconductivity. We are planning to calculate band structures for up to 10 000 materials using the Vienna ab-initio simulation package (VASP). For a low number of cores ($<16$) an almost linear speed-up of the calculations can be expected. To achieve a reasonable balance of computation time and performance we restrict the calculation of each structure to 16 cores. Each computation takes up to 24 hours of computation time meaning 384 core hours by using about 10 Gb of memory. Considering the number of organic materials we shall include in our calculations, we request a medium allocation. The output of the calculations will be transfered regularly to keep the storage demand low ($< 500$ Gb). Due to the available VASP installation as well as the large number of usable cores the system Triolith at the National Supercomputer Centre at Linköping University would be fitting well to our ideas. Full text can be found at: