Surface Adsorption

© 2017 Elsevier Inc. All rights reserved. Practical aspects are considered that users need to know in choosing computational methods and conceptual ideas for understanding surface adsorption embodying some degree of van der Waals dispersive attractions. Emphasized is the unique chemistry that results when these forces are strong and mix with traditional covalent or ionic bonding mechanisms to control chemical outcomes. Indeed, many methods are being developed to provide accurate descriptions of surface adsorption, including those based on ab initio and density-functional-theory (DFT) approaches. Accurate methods like coupled-cluster theory (CC), the random-phase approximation (RPA), and time-dependent DFT (TDDFT) are demanded when the critical interactions are weak, but these methods are computationally expensive, scaling formally at least as order n6 in system size (though efficient modern algorithms can reduce that to, e.g., order n4logn or even to linear scaling). How such methods can be applied to large systems is discussed, but more focus is given to more approximate, faster approaches, especially: the empirical pairwise additive D3 correction of Grimme et al., the van der Waals density functional approach of Dion et al. (vdW-DF), and the new semiempirical many-body dispersion (MBD) method of Tkatchenko et al. Fundamental understanding of all methods is provided by describing the way they behave in the very short range (unified atom) and very long range asymptotic limits, using Dobson's classification scheme. How simple computational methods can give realistic adsorption energies and structures for close-contact interactions whilst providing very poor descriptions of these asymptotic limits is examined. Four strong-interaction test systems are considered: the adsorption of benzene on coinage metals, the adsorption of 1,10-phenanthroline (PHEN) on Au(111), the adsorption of large tetraalkylporphyrin molecules on graphite from organic solution, and the nature of the Au-S bonds found on gold surfaces and nanoparticles protected by sulfur compounds. For adsorption energies, the vdW-DF method is found to be cumbersome and unreliable, the MBD method to be accurate but often prohibitively expensive, and the D3 method to be computationally efficient and, for the considered examples, highly reliable. Alternatively, neglect of dispersion interactions is found to lead to poor understanding of all important chemical properties