Fig. 1: Computational screening for CO2/water selectivity using isolated molecular building blocks.
From: Bottom-up computational design of shape-selective organic macrocycles for humid CO2 capture
a, Plot showing the Pareto front of calculated CO2 binding energy, \(E_{\rm{CO}_2}\), versus CO2 against water, \(E_{\rm{CO}_2-\rm{H}_2\rm{O}}\), for molecules in the CHN dataset (9,614 molecules in total, see Supplementary Fig. 2 for plot of full dataset; only data points with \(E_{\rm{CO}_2-\rm{H}_2\rm{O}}\) < 0.40 eV are shown here). Note that chemisorptive amidines are essentially outliers, and not our focus here, so we have defined the Pareto front by non-amidine molecules. The highlighted molecular structures can be clustered into four groups: group 1, amidines; group 2, amines; group 3, aromatic molecular clips featuring tweezer-like aromatic regions at obtuse angles to one another. Two of these molecules are presented as 3D structures to illustrate the clip geometry. Grey, carbon; blue, nitrogen; hydrogen atoms are omitted for clarity; group 4, aromatic compounds. The shaded colourful boxes (red, orange, blue and cyan) represent how the molecules in groups 1, 2, 3 and 4 interact with CO2. b, 1,694 extra data points were added to the original CHN dataset by including additional molecules with high molecular similarity to those in groups 1–3. The grey line is the original Pareto front and is same as the Pareto front in Fig. 1a above. The blue line is the new Pareto front after the addition of this new dataset; the structures that advance the Pareto front are all clip-like (group 3). Triangles represent (+)-NDI-Δ, whereas square represent P4.
