Machine Learning and Experimentally Exploring the Controversial Role of Nitrogen in CO2 Uptake by Waste-Derived Nitrogen-Containing Porous Carbons

Waste-derived nitrogen-containing porous carbons were widely accepted as promising carbon capture materials. However, roles of nitrogen in CO2 uptake were highly controversial, posing a challenge in designing high CO2 uptake porous carbons. Herein, nitrogen-containing species was firstly introduced into machine learning (ML) models to uncover the complex relationship of nitrogen, micropore and CO2 uptake by combining ML models, DFT computations and experiments. The results revealed that micropore volume (V-micro) was the most important property influencing CO2 uptake, but was not the only determinant factor. Nitrogen-containing species (pyrrolic/pyridonic-N (N-5) and pyridinic-N (N-6)) rather than total nitrogen content, also played an essential role. On the one hand, they can enhanced CO2 adsorption by Lewis acid-base and hydrogen bonding. On the other hand, they promoted development of micropores by participating in activation reactions. The model further indicated that excessive N-5 (>1.5 wt%) or N-6 (>1.7 wt%) led to restriction on developments of micropores, which was attributed to enlargement of pore size, collapses or blockage of micropores. The double edged-sword effect of N-5 and N-6 on changes of microporous structures was responsible for the long-standing controversy over nitrogen. The result was further verified by synthesizing eight porous carbons with different textural and chemical properties. This study provided not only a new perspective for resolving the controversy of nitrogen in CO2 uptake, but also a graphical user interface prediction software meaningful for designing porous carbons.