[{"type":"speech","title":"Multiscale material modelling on high performance computer architectures","issued":{"date-parts":[["2013"]]},"author":[{"family":"Danilov","given":"D."},{"family":"Meded","given":"V."},{"family":"Kondov","given":"I."},{"family":"Wenzel","given":"W."}],"note":"International Conference on Computational Modelling of Nanostructured Materials (ICCMNM 2013), Frankfurt, September 3-6, 2013","abstract":"112\n|\n'Institute of Nanotechnology, Karlsruhe Institute of Technology,\n2Steinbuch Center for Computing, Karlsruhe Institute of Technology,\nWith the accelerating materials development cycles, the development of Simulation\napproaches for predictive, de-novo characterization and optimization of materials and device\nproperties emerges as a grand challenge. A unified multi-disciplinary approach towards the\ndeployment of models, tools, algorithms and Simulation and visualization techniques is required to\ntransform isolated solutions for specific problems into comprehensive, industry-ready platforms,\nwhich are capable of predicting the properties of complex materials on the basis of their constitutive\nelements. While many techniques exist to address the specific questions, a lack of Integration of the\nexisting methods into readily available multi-scale modelling platforms has to date limited the\nimpact of materials-modelling techniques in materials design.\nOur concept will be illustrated for one application, modelling of charge transport through\norganic light emitting diodes (see Figure I), in order to demonstrate the strengths of our approach.\nAt the macroscopic level, an organic light emitting diode is a multi-layer device comprising many\ndifferent materials, each of which performs some key function in the device. The individual\nmaterials must be integrated into an electronics device and one should understand their properties as\na whole in order to optimize charge transport and light emission. That is, device Performance is a\ncharacteristic property not of an individual component but of the whole system, and can be\ndescribed by a set of partial differential equations governing charge transport and recombination.\nThe parameters of these partial differential equations depend on the properties of the individual\nlayers of the device, the nature of their Interfaces and even the process by which it has been\nassembled. All of these properties can be only understood at the quantum level, where materials\nsimulations can be performed only for fractions of the overall device that cover less than 0.001% of\nits total volume.\nIn order to model the f\u00fcll device, we must therefore define a workflow which links existing\nSimulation methods for each of the length and time scales required to treat this problem. At the\nlowest level this requires 100,0000s quantum calculations to characterize the hopping processes of\nelectrons between individual molecules in a small sample of the material. The parameters gathered\nat this level are then fed into atomistic Simulation methods, which in tum generate microscopic\nparameters characterizing an individual layer or an Interface. The latter parameters can then be used\nat the continuum level to characterize the whole device. The multi-scale modeling opens new\nopportunity to accelerate the materials-research development cycle by in-silico experiments. It will\nenable Virtual prescreening of OLED materials and prototype development of OLED devices in\norder to guide and support traditional experimentation.\nmacroscopic system molecular modelling quantum mechanics\nFigure 1: OLED simulations require in addition to tlie three scales sketched here,\nSimulation modules to generate the morphologies and calculate charge-transport\nproperties. A single Simulation thus requires Integration of five complex, but existing\nSoftware solutions.","kit-publication-id":"230091978"}]