In a first-of-its-kind initiative, researchers are using artificial intelligence (AI) to sift through more than 32 million possible materials for new batteries. This effort seeks to find alternatives that would be safer, cheaper, and better for the environment. From 2023 to 2024, scientists from UC Davis, TNC, and the U.S. Combined, they represent the promise AI holds to transform materials science and chemistry as we know them.
This effort relies on an AI model that was trained on quantum-accurate data. This enables it to perform rapid predictions of materials, and at only 1% the cost of classical computing. For chemists, AI can help them simulate a given molecule’s properties with incredible precision. It is extremely computationally expensive, restricting simulations to molecules with just a couple hundred atoms. Recent advancements are shifting this paradigm.
The Role of Electron Correlation in Material Discovery
Electron correlation plays an essential role in the physics of systems with strong electron correlations. This is most easily observed in systems that display topologically distinct electronic states. This continual interaction is perhaps the most important aspect of informing our understanding of how different materials perform under different conditions. By taking electron correlation into account in their models, researchers are now able to remove this bottleneck and in doing so dramatically increase the accuracy of their simulations.
As this technology develops, AI might be able to assist scientists in triaging long lists of possible materials. The AI model then expertly sifted through that pool of 500,000 candidates. Then it further winnowed them to 800 of the most promising options to test on a larger scale. This new capability is transformative because it has enabled researchers to target their laboratory efforts to the most promising candidates.
AI’s enhanced capacity to pinpoint “first-time right” candidates saves time and cuts costs when pursuing research. Now, researchers can narrow their efforts and select the most promising molecules to home in on. This enables them to take up only these selected candidates to the laboratory for further synthesis and testing, excluding surplus materials. This efficiency saves time as well as the expenses tied to material and discovery costs.
Collaboration and Future Prospects
Our collaboration with PNNL is a great example of how advanced and novel AI models can help to streamline the evaluation of materials. Researchers are increasingly recognizing the significance of combining AI with quantum computing as a means to address complex challenges in materials science. The combined power of these technologies holds the potential to revolutionize every discipline’s scientific research.
We believe that quantum-enhanced AI will accelerate our collective efforts to tackle the world’s greatest challenges. With this new technology, we will take big steps toward creatively fighting climate change and disease. If the pace of development can be maintained, specialists expect that efficient chemistry simulations that are currently beyond the capability of classical computation will soon be possible. Reliable quantum computing will require fault tolerance by encoding quantum information redundantly in logical qubits.
Experts estimate that achieving reliable quantum computing will require hundreds to thousands of high-quality qubits, with error rates around 10^-15. Overall, scientists expect on the order of a million physical qubits will be necessary to achieve this aspiration. As these changes shake out, the impacts on core scientific research and technological innovation start to get pretty profound.
Flattening the Curve of Materials Discovery
AI improves precision, but it levels the playing field in the race for materials discovery. The more clearly they see the big-picture dynamics, the faster and better researchers can predict similar systems. This capability greatly speeds the discovery and screening of new materials. That’s an especially important capability in rapid-moving sectors. Battery technology and renewable energy don’t wait on these big, timely wins.
With quantum computing fast approaching, the precise effects of AI are beginning to take shape, allowing it to reshape our material science future. All of these technologies, used in conjunction, are helping us gain a greater understanding of these complex systems. In turn, we create more durable and efficient materials applicable to not just transportation infrastructure.