AI for Sustainability: Materials Informatics is Driving Change, Finds IDTechEx
Materials informatics is the application of data-driven methods, including machine learning techniques, to the field of materials science.
This important part of the digital transformation toolbox for the materials industry has wide-ranging benefits, but the ability to enhance the sustainability of materials and their manufacturing processes could be the most impactful of these.
While interviewing materials informatics industry insiders for its report, “Materials Informatics 2023-2033”, IDTechEx found that decarbonization efforts are a growing driver for adopting these technologies and processes.
This, alongside the fact that materials informatics helps organizations to save money while accelerating materials innovation, is a contributing factor to IDTechEx’s prediction that the market for the provision of external materials informatics services will grow at 13.7% CAGR to 2033.
Players from across the AI industry are seeing materials informatics’ ability to contribute in solving the climate crisis.
Meta AI (of Facebook parent Meta) and Carnegie Mellon University’s Open Catalyst Project aims to identify catalysts that aid the production of fuels using excess renewable energy.
This project open-sources the discovery process, making the results of 260 million density functional theory (DFT) calculations publicly available for researchers to train their own surrogate models on.
Alongside the project’s initiators, universities, including Munich Technical University and other AI giants, including Tencent AI, have published results calculated from the dataset.
Applications of “AI for good” in sustainability of this sort will likely become a major part of the ESG toolboxes of machine learning industry titans. These surrogate models could aid in decreasing the energy requirements to produce, for example, green hydrogen.
IDTechEx covers the growing green hydrogen industry in its recent report, “Green Hydrogen Production: Electrolyzer Markets 2023-2033”.
Solar photovoltaics (PV) are another fruitful area for materials informatics to make an impact.
The IDTechEx report, “Materials Informatics 2023-2033”, discusses nine projects using machine learning to speed up the development of new PVs, including an MIT study using active learning to identify lead-free perovskites and work from Osaka University to produce correlations for bandgap and other properties from the structures of organic PVs.
However, AI can facilitate yet more areas of PV development, including accelerated lifetime testing.
A March 2023 working paper from researchers at institutions including MIT and Microsoft details a model called DeepDeg, which its authors say speeds up degradation testing of organic solar cells by up to twenty times.
Crucially, DeepDeg is designed to explain the degradation process of the candidate material, offering greater value to a selection process than simply predicting its lifespan.
The involvement of Microsoft shows yet again the growing interest of the best-known AI players in materials informatics, particularly for sustainability projects.
Established firms are not alone in recognizing the potential of materials informatics to facilitate decarbonization. Early-stage firm ExoMatter is a spinout of the German Aerospace Center that emphasizes sustainability in its materials identification platform.
It told IDTechEx that its platform not only can predict the carbon impact of the material candidates it identifies but had been used to identify 90 inorganic adsorbent candidates with better carbon retention properties for Dutch direct air capture startup Carbyon.
Carbyon estimated this had saved it at least six months in the lab.
The materials industry itself is acting on the need for in-house data science expertise as its importance continues to grow, including in facilitating sustainable manufacturing.
In February 2023, materials industry giant Toray Industries announced that it would be opening a new research facility for just this purpose.
The plan is to bring together materials scientists, chemical engineers and digital transformation professionals to drive nanotechnology advances using materials informatics and computational chemistry.
This highlights the increasingly cross-disciplinary nature of materials informatics, which has the potential to reduce the environmental impact of materials across their entire lifecycle, including their manufacturing processes and eventual disposal.