Exploring Metal Cluster Catalysts Using Swarm Intelligence: Start with Hydrogen Adsorption

Exploring Metal Nanocluster Catalysts for Ammonia Synthesis Using Informatics Methods: A Concerted Effort of Bayesian Optimization, Swarm Intelligence, and First-Principles Computation

“…The need for a distance greater than 15 Å is based on the literature. 67 − 69 The results of the literature survey, 128 − 130 which shows that a vacuum layer thicker than 15 Å is generally introduced when building a slab model, also support the validity of setting this distance. We refer to previous studies for the calculation of clusters using such a method.…”

“…We thought we would save a little bit of computational resources. We had already determined the structures of the Fe 5 , Ni 5 , and Cu 5 clusters in our previous work . So, we used those structures to evaluate E ads and E des .…”

“…We had already determined the structures of the Fe 5 , Ni 5 , and Cu 5 clusters in our previous work. 69 So, we used those structures to evaluate E ads and E des . We determined that the number of metal nanoclusters comprising the initial data set was 5, so we needed 2 more.…”

“…The need for a distance greater than 15 Å is based on the literature. − The results of the literature survey, − which shows that a vacuum layer thicker than 15 Å is generally introduced when building a slab model, also support the validity of setting this distance. We refer to previous studies for the calculation of clusters using such a method. − ,− In the VASP calculation, the Perdew–Burke–Ernzerhof (PBE) functional, a generalized gradient approximation (GGA), was used. Grimme’s D2 dispersion correction was applied .…”

“…The objective function is usually defined as the total energy of the system, calculated using quantum mechanical methods such as DFT. We have been using the PSO algorithm to search for cluster structures. − …”