How Google DeepMind's Al breakthrough could revolutionise chip, battery development
- Google DeepMind researchers recently unveiled a breakthrough in materials discovery using AI called Graph Networks for Materials Exploration (GNoME).
AI's Role in Predicting Structures
- GNoME utilized AI to predict the structures of over 2 million new materials.
- The potential applications extend to renewable energy, battery research, semiconductor design and computing efficiency.
Significance of the Breakthrough
- GNoME increases the number of 'stable materials' known to humanity by tenfold.
- These materials include inorganic crystals that modern tech applications from computer chips to batteries rely on.
- To enable new technologies, crystals must be stable otherwise they can simply decompose.
- However, these materials will still need to undergo the process of synthesis and testing.
Output and Filtering
- GNoME identified 381,000 of the 2.2 million crystal structures as most stable.
- The breakthrough aids ongoing research, such as finding .
- Stable solid electrolytes for Li-ion batteries.
- New layered compounds similar to graphene.
How does GNoME actually work?
- GNoME is a graph neural network model using active learning to scale up its dataset.
- This makes the algorithm “well suited” to the science of discovering new materials, which requires. searching for patterns not found in the original dataset.
Training Data and Collaboration
- GNoME was trained on crystal structure data from The Materials Project, a collaborative initiative providing data for inorganic materials research.
- The model underwent repeated assessments using Density Functional Theory (DFT) to understand atomic structures and crystal stability.
Precision Improvement
- GNoME assesses structural and compositional pipelines, evaluating candidates with known crystal structures and based on chemical formulas.
- Its precision rate for predicting material stability increased from 50% to around 80%.
- It claims equivalent knowledge to nearly 800 years of traditional computational methods, with 380,000 stable predictions available for further research.
Potential Impact on Materials Discovery
- The breakthrough accelerates the materials discovery process by filtering potential candidates.
- This enables researchers to focus on synthesizing materials with specific properties.