Table 8 GNN predictions for material Properties.

From: An integrated AI-driven framework for maximizing the efficiency of heterostructured nanomaterials in photocatalytic hydrogen production

Input features

Predicted bandgap energy (eV)

Predicted defect density (per cm3)

Experimental bandgap energy (eV)

Error (%)

TiO2-ZnO (1:1 ratio), 2.5 Å bond

2.8

1.2 × 10151.2 \times 10’{15}1.2 × 1015

2.85

1.75

TiO2-ZnO (1:2 ratio), 2.7 Å bond

2.9

1.5 × 10151.5 \times 10’{15}1.5 × 1015

2.87

1.05

TiO2-ZnO (2:1 ratio), 2.6 Å bond

2.7

1.1 × 10151.1 \times 10’{15}1.1 × 1015

2.75

1.81