Simulation of Metal-Supported Metal-Nanoislands: A Comparison of DFT Methods.

We have evaluated various density functional theory (DFT) methods to simulate geometric, energetic, electronic, and hydrogen adsorption properties of metal-nanoparticles supported on metal surfaces. We used Pt and Pd nanoislands on Au(111) as model systems. The evaluated DFT methods include GGA (PW91, PBE, RPBE, revPBE, and PBESol), GGA with van der Waals (vdW) corrected (PBE-D3), GGA with optimized vdW functionals (revPBE-vdW), meta-GGA (SCAN and MS2), and the machine learning-based method BEEF-vdW. The results show that the various DFT methods yield similar geometric and electronic properties for Pt (or Pd) nanoislands on Au(111). The DFT methods also produce similar relative energetics for small Pt (or Pd) clusters with different conformations on Au(111). The results show that a triatomic cluster of Pt on Au(111) is more stable with a linear conformation. In contrast, a triatomic cluster of Pd is more stable with a triangular conformation. For clusters with four or more atoms, Pt and Pd clusters on Au(111) prefer non-linear conformation. We found that the various DFT methods yield different results only for the adsorption energy of hydrogen.