Please note: unless otherwise specified, the internships are only available for students with a nationality of an EU-member state and/or students from a Dutch university.
DIFFER (Dutch Institute for Fundamental Energy Research) is one of the Netherlands Organisation for Scientific Research (NWO) institutes and focuses on a multidisciplinary approach to energy research, combining physics, chemistry, engineering and materials science. The institute is based on two main strands, solar fuels for the conversion and storage of renewable energy and fusion-energy as clean and unlimited source of energy. DIFFER is developing and supporting a national network on fundamental energy research and is closely collaborating with academic institutions, research institutes and industry. The institute is located in a new building at the campus of Eindhoven University of Technology (TU/e).
The discovery of new energy materials is becoming a large-scale challenge that is far beyond the reach of experimentation but also stretching the limits of conventional computation. At DIFFER; our AMD research group is working on to improve the speed and the prediction capability of computational methods for the discovery of new energy materials. We use machine learning (ML), which is essentially a method to make predictions and to optimize a performance criterion based on the available example data.
In this project, we plan to use existing ML methods to learn on the quantum mechanically computed chemical data of compounds that is generated through the calculations of their reactions on the catalytic surfaces of metals and alloys. The pre-processed and curated data contains about a few thousand data instances in a CSV text format and is readily available in our group. We plan to use this data as a starting point (i.e. training set) of a machine-learning model. The primary task of the MSC project is to learn on the hidden correlations between the different features (i.e. properties of materials). In other words, the aim of the project is to demonstrate the composition-structure-property relationships of the materials that are represented in our dataset that may not be immediately visible to human researchers.
The work requires familiarity with Python language and experience with the use of functions, classes, methods, and lists. Prior experience with any ML method and/or knowledge of electronic structure, such as DFT, calculations will be helpful, but not required. This is an ambitious project with room for new coding and ML skills development through professional guidance of our group members. Though the project is actually computer science centered, we plan to utilize the research outcomes of your project to intensify our research group efforts in a societally rewarding and scientifically challenging research topic of CO2 conversion.
Intended level: MSC, ideally a 9-12 months project
Expertise needed: computer science
Please apply via email including your CV to: s  er  differ  nl
Go to the Vacancies page.