Google’s DeepMind uses machine learning to predict and synthesize novel crystal compounds
Crystals have long fascinated scientists and the general public alike, with their mesmerizing beauty and unique properties. Now, Google’s artificial intelligence company, DeepMind, has taken the study of crystals to new heights. In a groundbreaking research effort, DeepMind has developed a machine-learning tool called GN o ME (Graph Networks for Materials Exploration) that can predict and generate new crystal structures. This technological breakthrough has the potential to revolutionize various industries, from renewable energy to electronics.
Exploring the Vast World of Crystals
Crystals are a diverse family of compounds, characterized by their repeating atomic structures. With approximately 48,000 known crystal structures, scientists have only scratched the surface of their potential applications. DeepMind’s GN o ME has expanded this knowledge exponentially by predicting 2.2 million new crystal structures, previously unknown to science.
Validating the Machine’s Predictions
To validate the accuracy of GN o ME’s predictions, DeepMind collaborated with researchers at the University of California, Berkeley. They successfully synthesized 41 out of 58 compounds chosen from the machine’s predictions in just over two weeks. Furthermore, since the preparation of their paper, other research groups have produced over 700 additional crystals based on DeepMind’s work.
Unleashing the Potential of New Crystal Structures
DeepMind’s contribution to crystal research extends beyond prediction. The company has made a subset of what they believe to be the 381,000 most stable crystal structures publicly available. Among these structures are thousands with potential for superconductivity, where electrical currents flow without resistance, and hundreds that could serve as conductors of lithium ions, crucial for battery technology. This vast expansion of candidate materials presents researchers with a wealth of possibilities for further exploration.
The Beginning of a New Era
While DeepMind’s achievement is undoubtedly impressive, it is only the beginning of a vast exploration of crystal structures. Materials scientist Aron Walsh from Imperial College London emphasizes that this research has merely scratched the surface of what is possible. His own calculations suggest that there could be as many as 32 trillion stable crystals incorporating four chemical elements. Additionally, GN o ME’s focus on low-temperature and low-pressure crystal formation means there is still much to learn about crystals that form under different conditions.
Beyond Crystal Structures
While DeepMind’s work has primarily focused on crystals, the potential applications of AI in materials research extend far beyond this realm. The techniques used to predict crystal structures could pave the way for discoveries of unknown rules governing their formation. Furthermore, the time-consuming process of synthesizing and testing new materials could be expedited by AI, allowing scientists to explore the behavior of the 2.2 million new crystal structures without the need for manual synthesis.
Conclusion:
DeepMind’s AI breakthrough in predicting and generating new crystal structures has opened up a world of possibilities for scientists and industries alike. With 2.2 million new crystal structures to explore, researchers can now delve into uncharted territory in fields such as renewable energy, electronics, and beyond. While the practical applications of these new structures remain to be seen, the knowledge gained from this research and the potential for further discoveries are invaluable. As AI continues to advance, the boundaries of materials science are being pushed further, unlocking the secrets of the natural world and propelling innovation to new heights.
Leave a Reply