Size Effects on the Electronic and Mechanical Properties of Diamond‐XII: A Machine Learning Potential Study

Ultrahard carbon materials offer great potential for energy storage and advanced devices. Diamond‐XII, a cubic carbon allotrope with twelvefold diamond segments, stands out for its stability and hardness. This study refines a deep potential model for Diamond‐XII using potential fine‐tuning and first‐principles calculations. Structures with n = 1–11 are optimized, showing a turning point in energy and density at n = 9. All exhibit Vickers hardness above 50 GPa, with decreasing bandgap from n = 1 to 9. The results highlight the power of deep learning potentials in designing ultrahard carbon materials.