Machine learning could be used to optimize carbon dioxide absorption

Machine learning could be used to optimize carbon dioxide absorption


Carbon dioxide is a compound related to the greenhouse effect and belonging to the so-called greenhouse gases (GHG), associated with global warming and climate change. In 2021, the National Oceanic and Atmospheric Administration of the United States indicated that a new historical level of carbon dioxide in the atmosphere was reached, 50% higher than pre-industrial levels. For this reason, there is a concern to reduce the levels of this GHG from different fronts, among them, that of industrial carbon dioxide. In this area, machine learning could be used to optimize the absorption of carbon dioxide capture and storage materials.

Due to the shortcomings of current carbon dioxide capture and storage methods, researchers are working to develop what could be the next generation of these materials. Through the use of agricultural, food, animal or forest waste, the creation of porous carbons derived from biomass residues (CPDRB) is proposed. Using these porous materials for the absorption of carbon dioxide in industrial afterburner sources, it is considered, would be an important step towards a circular economy, by taking advantage of waste. However, given their novel emergence, there is still no scientific consensus as to how CPDRBs should be synthesized or what properties and compositions should be achieved in their manufacture.

For this, a study published in the journal Environmental Science & Technology turned to machine learning to solve this question. Based on the review of research on CPDRB, they trained their machine learning model which, based on three central factors, is responsible for determining the absorption of carbon dioxide from a given CPDRB. According to the results of the machine learning model, the absorption parameters at which a CPDRB operates, such as temperature and pressure, turned out to be the most important central factor. This underscores the importance of optimizing the operating conditions of the material first, the study authors concluded. The CPDRB’s textural properties, such as pore size and surface area, ranked second and their elemental composition last.

The researchers note that the predictions of the machine learning model support the existing literature on the subject and current understanding of carbon dioxide capture mechanisms. In addition, they point out, the applicability of this machine learning model could be extended to other types of materials beyond CPDRB.

In addition to this, other investigations regarding the absorption of carbon dioxide promise important advances in the matter. Chemists at the University of Bayreuth in Germany previously unveiled the development of a material capable of specifically separating carbon dioxide from industrial waste gases. A mixture of organic and inorganic compounds, this material differs from separation processes by completely removing carbon dioxide from gas mixtures without chemically binding it. In addition, according to the researchers, it can be used with the mixture of waste gases from industrial plants, but also with natural gas or biogas.

At this point, global warming portends many catastrophic changes around the globe, many of them already irreversible. In order not to aggravate the situation and reverse what is possible, it is essential to create mechanisms to reduce carbon dioxide emissions. As in many other areas of human development, everything points to a crucial role for technologies, such as machine learning, in the optimization of carbon dioxide absorption materials in the industrial field.