Fig. 5: Physicochemical insights from interpretable deep learning. | Nature Communications

Fig. 5: Physicochemical insights from interpretable deep learning.

From: Interpretable design of Ir-free trimetallic electrocatalysts for ammonia oxidation with graph neural networks

Fig. 5

a Local interpretation of Nh adsorption on Pt3Ru(100) relative to Pt(100) and Pt3Ru1/2Co1/2(100) to Pt3Ru(100). b A graphical illustration of the Newns-Anderson model for an adsorbate resonance state interacting with metal d-states. DFT-calculated DOS projected onto the 2pz orbital of *Nh on Pt3Ru1/2Co1/2, Pt3Ru, and Pt are shown with adsorbate-metal antibonding states highlighted for comparison. c Global SHAP analysis of TinNet models of Nb adsorption showing a strong correlation of the machine-learned adsorbate resonance energy with the adsorption-energy contribution, i.e., the SHAP value.

Back to article page