AI Accelerates Discovery of Next-Gen Carbon Capture Materials

Researchers have been searching for efficient, affordable, and sustainable methods to capture carbon dioxide emissions. Existing CO2 absorption technologies that rely on chemical reactions have several drawbacks.

Metal-organic frameworks (MOFs) have emerged as a promising alternative for CO2 capture, garnering significant attention in scientific research over the past two decades.

A MOF (metal-organic framework) is a material made by combining organic molecules and metal ions or groups, where the metals act as connectors and the organic compounds act as linkers.

MOFs are an atomic-scale version of magnetic ball-and-stick toys. These materials come in various shapes and sizes, from 1D to 3D, with tiny pores that make them highly porous. MOFs are known for their crystal-like structure, large surface area, high-temperature stability, and ability to adjust the size and shape of their pores. These features make MOFs effective for removing different types of pollutants over long periods.

When metal clusters and linkers are mixed, typically in a liquid, they naturally self-assemble into a pore system that normally remains stable even after the solvent or other molecules inside the pores are removed. The combination of metal and organic materials in MOFs gives them unique properties.

There is a vast variety of MOF structures, each with unique properties, such as their ability to store or absorb a specific gas and release it. The possibilities are endless, with over 90,000 different MOFs already discovered and many more yet to be found. One of the most well-known MOFs is MOF-5, which has an extremely high surface area of 2200 m²/cm³, about 15 times that of human lungs.

Gases are unique in two main ways: their stickiness (or polarisability) and their size. Researchers design and select MOF materials based on these properties to create the best method for gas storage.

By adjusting the size and properties of the pores, they can design an MOF to store one gas while selectively ignoring another. However, the efficiency of a MOF in storing a particular gas, like carbon dioxide, can also be influenced by its environment—for example, whether it’s located in a humid sea-level area or a dry, high-altitude region.

AI helps identify and select the best MOF for a specific application

Designing an MOF for a particular gas and its optimum function and efficiency, including cost efficiency, will require sorting through the billions of possibilities to find the right MOF. This is almost impossible for a human chemist, but a perfect challenge for an artificial intelligence (AI) model.

According to The Economist, AI startups are developing systems to tackle this, with companies like CuspAI leading the way. By using multiple AI models working together, some are trained to generate candidate molecules with specific properties and assess their performance. This approach aims to create a system that can identify the best MOF for any environmental condition, demonstrating how AI can address challenges in materials science.

To demonstrate AI’s capabilities, a doctoral student in economics at the Massachusetts Institute of Technology (MIT) presented a paper at a conference in November 2024, analysing the impact of a new AI tool on materials research. The results were impressive: the number of materials discovered increased by 44%, the number of product prototypes using these new materials rose by 17%, and the number of patents filed jumped by 39%.

Using AI, the innovations that emerged appeared to be genuinely new. AI-assisted patents were more likely to include new technical terms, and the materials featured more unfamiliar physical structures. Other startups and research labs are racing to design and create the right MOF for the application.

But even with the combined help of AI and robotics, synthesising new materials for MOFs remains a challenge. Max Welling, a co-founder of CuspAI, warns that “Recipes are very finicky.” He explains that even small changes in humidity or air quality can derail a lab’s efforts to create the desired product.

While there are still challenges, there is reason for cautious optimism in MOF research. The success of AI startups in this field could lead to more efficient and cost-effective carbon-capture materials, showcasing AI’s valuable contribution to addressing climate change.

While the search for efficient, sustainable, and cost-effective methods to capture carbon dioxide continues, metal-organic frameworks (MOFs) present a promising solution with their remarkable properties and versatility. By combining AI and robotics, researchers can design and select the best MOFs for a specific application faster and more efficiently.

The progress seen in AI-driven MOF research, as demonstrated by startups like CuspAI, holds significant potential for creating optimised carbon-capture materials.

Although challenges remain, particularly in synthesising new materials under controlled conditions, the future of MOFs in addressing climate change looks promising with continued advancements in AI and material science.