Machine Learning Analysis Can Help Reduce Carbon Emissions

2019-10-23

A group of scientists led by Anatoly Frenkel from Stony Brook University has developed a new method that can help reduce carbon emissions by using artificial intelligence (AI) to convert carbon dioxide (CO2) into methane.


By tracking the size, structure, and chemical properties of catalytic particles under real reaction conditions, scientists can identify which characteristics correspond to optimal catalytic performance and use this information to guide the design of more efficient catalysts.


Frenkel said, "By converting carbon dioxide into methane, we have also created a sustainable non-fossil fuel energy source that not only reduces carbon emissions but is also easy to store and transport."Frenkel's team has been developing a machine learning approach to extract catalytic properties from X-ray features of chemical transformations. Based on X-ray data collected at DOE national laboratories, they described their findings in a paper published in the Journal of Chemical Physics.


Now, a team led by Stefan Vajda, a senior chemist at the J. Heyrovský Institute of Physical Chemistry, has conducted a series of experiments. They used mass spectrometry and X-ray studies at Argonne's Advanced Photon Source (APS) to investigate how clusters of various sizes behave in reactions and how their oxidation states evolve during the reaction of carbon dioxide with hydrogen.


Some oxides may enhance reactivity, while others may inhibit it. To understand how catalysts work, scientists need to know what types of oxides are formed during the reaction and how they affect catalytic performance.


Copper has been shown to be a catalyst that can lower the reaction temperature of carbon dioxide and methane. Clusters of various sizes of copper can also help drive the reaction to the desired outcome (selectively producing methane and water vapor) without guiding the reactants towards other products through various pathways.Frenkel said, "Broadly speaking, implementing this idea faces two main challenges. The first is a lack of understanding of the structure of prepared clusters; the smaller they are, the greater the potential for shape and structural changes, even with the same number of atoms in each cluster." The second is that even if we start the reaction with clusters of a certain size and shape, they may transform into various forms of oxides that are unrecognizable during the reaction."


It turns out that the most catalytically active state of copper catalysts is a mixture of metal clusters (where copper is only bonded to other copper atoms) and two different copper oxides (CuO and Cu2O). Frenkel's team is continuing their analysis to gain more information about the catalytic mechanism and will publish their results in the future.

RECENT AI NEWS

RECENT AI TOOLS