▎ 摘　　要

Locating the global minimum of a potential energy surfaceis anarduous task. The complexity of the potential energy surface increasesas the number of degrees of freedom of the system increases. The highlyrugged nature of the potential energy surface makes the minimizationof the total energy of the molecular clusters a difficult optimizationproblem. A solution to this conundrum is the use of metaheuristictechniques that efficiently track down the global minima through atrade-off between exploration and exploitation. Herein, we use theswarm intelligence technique, particle swarm optimization to locatethe global minima geometries of N-2 clusters of size 2-10,in free and adsorbed states. We have investigated the structures andenergetics of bare N-2 clusters, followed by N-2 clusters adsorbed on graphene and intercalated between the layersin bilayer graphene. The noncovalent interactions between dinitrogenmolecules are modeled using the Buckingham potential as well as theelectrostatic point charge model, while those of the N-2 molecules with the carbon atoms of graphene are modeled using theimproved Lennard-Jones potential. The interactions of the carbon atomsbelonging to different layers in a bilayer are modeled using the Lennard-Jonespotential. The bare cluster geometries and intermolecular interactionenergies obtained using particle swarm optimization are found to bethe same as reported in the literature, validating the use of particleswarm optimization for studying molecular clusters. The N-2 molecules are found to adsorb as a monolayer on top of the graphenesheet and intercalate themselves right in the middle of the two sheetsof bilayer graphene. Our study establishes that particle swarm optimizationis a feasible global optimization technique for performing the optimizationof high-dimensional molecular clusters, both in pristine and in confinedforms.