Understanding the Activity Origin of Pd-anchored Single-Atom Alloy Catalysts for NO-to-NH3 Conversion by DFT Studies and Machine Learning

Jieyu Liu,Junze Zhang,Haigang Deng,Shuoao Wang,Xingxing Jiang,Li Wang,Changhong Wang
DOI: https://doi.org/10.1016/j.cclet.2024.110656
IF: 9.1
2024-01-01
Chinese Chemical Letters
Abstract:The electrochemical nitric oxide reduction reaction (NORR) to NH3 represents a promising avenue for NO removal and NH3 synthesis. It is essential to develop catalysts with superior performance for this process. We systematically studied a series of single-atom alloy catalysts (SAACs) with Pd single-atom dopants using density functional theory (DFT) calculations and machine learning (ML). Based on the energetic span model, we take Gmax(η) as a descriptor to evaluate the reaction activity of SAACs. After comprehensively considering the stability, activity, and NH3 selectivity of SAACs, Cu and Pd/Cu SAAC are screened out as candidate NORR to NH3 catalysts. To predict the Gmax(η) descriptor, the extreme gradient boosting regression (XGBR) ML algorithm was adopted with geometric/electronic properties of the SAACs as input features. Additionally, we proposed a mathematical formula to correlate the crucial features and the Gmax(η) descriptor using the sure independence screening and sparsifying operator (SISSO) approach. This work provides an understanding of the complex NORR mechanisms and offers a strategy to rationally design highly efficient SAACs.
What problem does this paper attempt to address?