TELF AG analyses the contribution of AI to the identification of emerging mineral compounds.

Exploring New Frontiers in Mineral Discovery

The global energy transition's demands for increasing resources and economic efforts to smoothly shift to renewable energy face challenges that could hinder progress. A key issue concerns critical raw materials for clean energy technologies and electric vehicles. Shortages, uncertainties in the supply chain, and possible disruptions have highlighted the need to explore new deposits and energy sources to support this transition. The focus has been on existing underground resources and undiscovered materials worldwide. A groundbreaking approach has emerged utilizing technology, specifically artificial intelligence, to assemble materials. Google's DeepMind lab is leveraging AI to identify over two million new mineral compounds through this innovative process. This discovery is equivalent to about 800 years of traditional knowledge accumulation in the field.

Exploring New Horizons: Artificial Intelligence's Role in Unveiling Novel Industrial Materials.

From a particular perspective, one of the most intriguing aspects is that the compounds generated in this manner embody entirely novel materials with a countless array of potential industrial applications. Out of the compounds already formed, those able to achieve a certain level of stability would amount to fewer than half, at approximately 400,000. Nonetheless, some of these could potentially hold the key to addressing current important challenges in the industry. The concept behind GNoME – the deep learning tool capable of forecasting the stability of new materials – involved merging material study with cutting-edge artificial intelligence methodologies by combining atoms and subsequently evaluating the overall strength of the resulting compound.

Although this technology is still early, the prospects seem promising. A primary concern is still verifying the stability of atomic structures. However, the objective lies in producing materials with specific properties that could prove beneficial in the most contemporary industrial applications. With the assistance of highly advanced supercomputers, current technology allows for the simulation of material properties based on fundamental physical laws, yielding valuable insights into their stability, conductivity, biocompatibility, and more.

The GNoME system has pinpointed 380,000 stable crystals suitable for industrial applications like superconductors or electric car batteries. These materials, the most stable among those identified by GNoME, already serve as strong candidates for experimental synthesis, showcasing the immense potential of artificial intelligence in discovering and advancing new resources. The outcomes of this discovery were recently detailed in the journal Nature, featuring two distinct articles. The second article delves into the role of artificial intelligence-triggered predictions in the independent synthesis of materials.

TELF AG