“Revolutionary” materials and big data

| |
1 Star2 Stars3 Stars4 Stars5 Stars

Whether electric cars, smartphones or bio-implants – many technical innovations are based to a considerable extent on new materials. And finding such materials is an elaborate and largely evolutionary process. In future, modern big-data methods are expected to provide us with the necessary “Aha” moments.

Materials and constituents generally perform their duties entirely unregarded. Unjustifiably so! According to Germany’s Federal Ministry of Research, 70% of the gross national product of technologically active western countries is directly or indirectly associated with the development of new materials. Innovations in this area are therefore extremely important, but unfortunately can only be achieved with lengthy experimentation. For example, there are around 240,000 recorded inorganic materials, but we only know some of the properties of about 100 of them.

Max Planck researchers are hoping that their project “BigMax” will now change that. They plan to search existing measurement data relating to the properties of innumerable materials in order to detect patterns that will help identify the best material for a particular purpose.

“Exemplary” datasets

Many of the institutes of the Max Planck Society (MPG) already produce large volumes of data from experiments or computer simulations. For example, processes such as X-ray structure analysis and atomic tomography generate millions of data points per minute, from which researchers gain insight into the arrangement of atoms in a solid, for example. The quantum mechanical calculations used in solid-state physics and chemistry also generate enormous quantities of data, and these are already being used as the basis for deducing the properties of materials.

BigMax now wants to use new and improved methods to extract even more information from such big data. One core goal: to examine the datasets for particular structures or patterns. With the help of this entirely new information the researchers want to make it possible to theoretically predict the properties of metals and alloys and purposefully design polymer materials that have specific desired properties.

“Unexpected” materials with big data

We still have a long way to go before we can fulfill the dream of a multi-dimensional materials map in which we can simply look up which material is best for a particular purpose. In future, the task will be to apply patterns found in big data to materials that have not yet been analyzed at all.

This approach not only saves time and money, it can also far more easily lead to unconventional solutions, because experiments generally set out from established criteria. For example, someone who is looking for superconductors will primarily look at the class of substances in which such things have already been found. Not exactly the best place to find a “revolution”. By comparison, perhaps in future, data that was generated during research on photovoltaic cells could also lead to new thermoelectric materials.


Optimal material with big data (Image: MPG)

In future, it is hoped that big-data algorithms will identify the optimal material for a specific purpose. (Image: MPG).