Scientific workflows for accelerated computational materials discovery

With the prospect of an increase in global warming, clean energy materials helping to reduce emissions are needed more than ever. One of the cornerstones of accelerating the discovery of new materials with improved properties is the access to large amounts of high quality data obtained from systematic high throughput simulations and experiments. Despite the promises data-driven discovery holds, access to reliable data sets is often limited in the field of materials science. Scientific workflows in materials science help to create such reliable datasets by combining several subsequent computational tasks, which calculate specific material properties in an automated way. Moreover, they allow for a full provenance tracking and for ensuring reproducibility of the data. In this thesis two scientific workflows are implemented in the framework of Density Functional Theory (DFT) to accelerate the discovery of inorganic nanotubes and battery electrodes. The first workflow calculates stability properties of inorganic Janus nanotubes. Janus nanotubes consist of three asymmetric layers. The different elements inside and outside the tube lead to an embedded strain causing spontaneous self-rolling. Here, it is found that the preferred small radius of the tubes is influenced by the lattice mismatch of the corresponding parent sheets composing the Janus structure as well as a difference in bond-strength between the elements inside and outside the tube. The second workflow identifies ion insertion battery properties including thermodynamic properties, such as stability, volume change, and open-circuit voltages, together with kinetic properties, such as diffusion barriers. The workflow is tested for the specific example of Mg-ion batteries. In addition to a few candidates that have been identified for experimental verification, simple descriptors have been found, which can accelerate the exploration of new ionic conductors by predicting which materials and diffusion paths are worth to be calculated. While the workflows are able to produce reliable datasets, they have also proven to enable the sharing of data and knowledge in terms of best practices in between researchers. Through building on top of existing knowledge, it can be expected that the reuse and further extensions of the workflows can greatly improve the speed of computational materials discovery and pave the way for efficient data-driven discovery.