Georgia Tech and Meta have created a massive open-source database, named OpenDAC, to facilitate the design and implementation of direct air capture technologies. Direct air capture, a technology that removes carbon dioxide from ambient air, holds potential in addressing excessive carbon emissions and mitigating catastrophic climate impacts.
However, the unique environmental and location-specific requirements for each direct air capture system pose a significant challenge. The OpenDAC project aims to overcome this hurdle by providing a robust dataset that enables researchers to train AI models, accelerating the development of climate solutions.
The OpenDAC database, believed to be the largest and most robust of its kind, contains reaction data for 8,400 different materials and is powered by nearly 40 million quantum mechanics calculations. Georgia Tech researchers, led by Andrew J. Medford and David Sholl, provided the inputs for the database, including the structure of nearly every known metal-organic framework (MOF) and their interactions with carbon dioxide and water molecules. MOFs are a promising class of materials for direct air capture due to their cage like structure and proven ability to attract and trap carbon dioxide.
Meta's Fundamental AI Research (FAIR) team, led by Anuroop Sriram, generated the database by running quantum chemistry computations on the inputs provided by the Georgia Tech team. These calculations utilised approximately 400 million CPU hours, demonstrating the immense computational power required for such an endeavour. The FAIR team also trained machine learning models on the database, which, once trained on the 40 million calculations, could accurately predict how the thousands of MOFs would interact with carbon dioxide.
Using the OpenDAC database, the Georgia Tech and Meta teams identified approximately 241 MOFs with exceptionally high potential for direct air capture. These MOFs strongly attract carbon dioxide while not attracting other air components, such as water vapor. The identification of these high-potential MOFs is a crucial step in advancing direct air capture technology and bringing it closer to practical implementation.
The entire OpenDAC dataset project, including the data, models, and algorithms, is open-source, encouraging the scientific community to join the search for suitable materials. By making the project accessible to researchers worldwide, Georgia Tech and Meta aim to accelerate the development of negative-emission technologies like direct air capture, which are essential for achieving net-zero carbon dioxide emissions by 2050.