USC Viterbi scientists have made a groundbreaking advancement with Allegro-FM, an AI-driven simulation model that can predict molecular behavior of billions of atoms, offering a promising solution for creating carbon-neutral concrete and combating climate change.
In an era where climate change poses an immense threat to our planet, scientists at the USC Viterbi School of Engineering have unveiled a promising solution. They’ve developed an artificial intelligence model — Allegro-FM — that can simulate the behavior of billions of atoms simultaneously, potentially revolutionizing the design and production of materials like concrete.
The current reality of global warming is daunting, marked by relentless droughts, melting glaciers and increasingly severe storms and wildfires. Driving this environmental crisis is the incessant emission of carbon dioxide.
But a recent breakthrough from USC researchers offers a beacon of hope.
USC Viterbi professors Aiichiro Nakano and Ken-Ichi Nomura, leverage a longstanding collaboration of over two decades, to address these challenges and develop Allegro-FM.
This AI-driven model has made a significant theoretical discovery: the possibility of recapturing carbon dioxide emitted during concrete production and reincorporating it into the concrete itself.
“You can just put the CO₂ inside the concrete, and then that makes a carbon-neutral concrete,” Nakano, a USC Viterbi professor of computer science, physics and astronomy, and quantitative and computational biology, said in a news release.
Concrete production is a major polluter, contributing about 8% of global CO₂ emissions. By simulating various concrete chemistries virtually, Allegro-FM can potentially speed up the process of developing concrete that not only acts as a carbon sink but also exhibits improved mechanical properties.
The scalability of Allegro-FM is a crucial aspect of its innovation.
Traditional molecular simulations are limited to thousands or millions of atoms. In contrast, Allegro-FM has demonstrated extraordinary efficiency, simulating over 4 billion atoms with 97.5% efficiency on the Aurora supercomputer at Argonne National Laboratory.
This capability is approximately 1,000 times larger than conventional methods.
“Concrete is a very complex material. It consists of many elements and different phases and interfaces. So, traditionally, we didn’t have a way to simulate phenomena involving concrete material. But now we can use this Allegro-FM to simulate mechanical properties [and] structural properties,” added Nomura, a USC professor of physics and astronomy.
Concrete not only fires up the most extensive carbon emissions but also resists fire, making it a preferred construction material in wildfire-prone areas like Los Angeles.
The AI model suggests that carbon-neutral concrete could be a viable alternative in such environments, addressing both emission concerns and enhancing longevity.
Modern concrete has an average lifespan of about 100 years, whereas ancient Roman concrete has lasted for millennia. Nakano explained that by incorporating CO₂, the resultant “carbonate layer” could significantly extend the material’s durability.
“If you put in the CO₂, the so-called ‘carbonate layer,’ it becomes more robust,” Nakano added.
Behind these achievements lies the power of AI.
Traditionally, simulating atomic behavior required extensive and complex mathematical formulas grounded in quantum mechanics.
However, AI-driven processes have streamlined this with machine learning models, making the research faster and more technologically efficient.
“Now, because of this machine-learning AI breakthrough, instead of deriving all these quantum mechanics from scratch, researchers are taking [the] approach of generating a training set and then letting the machine learning model run,” added Nomura.
Allegro-FM’s potential extends beyond concrete, with applications in various fields by covering 89 chemical elements. The AI can simulate the interaction functions between atoms more accurately and efficiently than previous methods, which required separate equations for different elements.
Nakano and Nomura’s pioneering work, published in The Journal of Physical Chemistry Letters, signifies just the beginning of this revolutionary approach. They plan to continue refining their simulations to explore more complex geometries and surfaces.
Co-authors of the study include Priya Vashishta, a USC Viterbi professor of chemical engineering and materials science, and Rajiv Kalia, a USC professor of physics and astronomy.