Super-atom model

The development of the super-atom model for the description of electronic properties of metal clusters arose from the attempt to understand and interpret experimental data by W. Schulze and co-workers Figure 1.1 shows the absorption of small silver particles... 

Figure 1.1 Optical absorption of small silver particles Ag embedded in argon at low temperatures, according to Ref. [11]. The huge absorption hump is a collective electronic oscillation localized at the interface Ag/Ar. This figure historically gave the impact for the development of the super-atom model for metal clusters. For large clusters a broad, damped peak is observed, whereas for small clusters the line is fragmented...

Two features of the jellium description of the super-atom model are experimentally confirmed ... 

The Co Q cluster was investigated using the same -super-molecule model as was used for Co. With ten d" atoms, the optimal cluster can be described as two pyrcunids in their lowest quartet states bound together by their square bases. Hence, a triple-bonded species (one cT bond and two bonds )... 

In a meso-scale model, a group of atoms is often replaced by a single interaction center. This center is usually the size of a monomer in a polymer and it is often called a super-atom. Because the super-atoms are the only interaction centers in a meso-scale simulation, they are required to carry the information of the interactions between the real atoms in their local geometrical arrangements that are imposed by the chemistry of the polymer. 

Figure 2 Various anchor sites for the super-atoms of polystyrene. Reprinted from Computers and Chemical Engineering Volume 29, Q. Sun and Roland Faller, Systematic Coarse-Graining of Atomistic Models for Simulation of Polymeric Systems, pp. 2380-2385, Copyright (2005) with permission from Elsevier.

Coarse-graining techniques are based on the idea of effective interaction potentials between super-atoms. The larger-scale model is calibrated against the smaller-scale model, which is considered to be more reliable and is used as a gold standard. Thus, any deficiencies of the small-scale model are carried over to the large-scale model. 

In principle, the mapping onto simple models assigns a computationally cheap interaction potential to a set of super-atoms. With respect to optimization, this mapping is just the initial step before any further refinement is carried out. In this vein, the polymer models can be similar to one another, in which case, we can get a good mapping or we can rely on vastly different polymer models with poor mapping qualities and, as such, have essentially nothing to do with each other. 

Because we do not have to follow a physical trajectory in an MC simulation, we can also use models that are further removed from the true physical or chemical reality. Such models include lattice models (see, e.g.. Refs. 28,61,62). With lattice models, the space of our system is (typically) evenly divided into cells, each of which are represented by one lattice site. Lattices can be very simple cubes or they can be specially adapted, highly connected grids. Here again, we need super-atoms, which, however, can occupy only lattice sites. In most lattice models every site is either singly occupied or empty, meaning that the interaction sites have an impenetrable hard core, which contrasts to Lattice-Boltzmann models used in studies of hydrodynamics in which every lattice site is occupied by a density, in which case, one deals with a density-based field theory. In lattice models, there exist only a fixed number of distances that can be realized. It makes no sense to distinguish between, say, a... 

In most lattice models, a super-atom can represent a monomer or a Kuhn segment of the chain.In most lattice models, only interactions of very close neighbors (first or second neighbors) are included such that the calculation of the energy, which is the computationally most expensive part of a Monte Carlo calculation, is a sum whose calculation scales linearly with the number of lattice sites. The actual mapping process, if done systematically, is easier than without using a lattice. With a lattice model, we have fewer points in the RDF that need to be reproduced otherwise, there is no fundamental difference between lattice and off-lattice models. 

At the end of last century, a near frictionless carbon (NFC) coating was reported, which is practically hydrogen contained DLC film grown on steel and sapphire substrates using a plasma enhanced chemical vapor deposition (PECVD) system [50]. By using a ball on a disk tribo-meter, a super low friction coefficient of 0.001-0.003 between the films coated on both the ball and the disk was achieved [50]. A mechanistic model was proposed that carbon atoms on the surface are partially di-hydrogenated, resulting in the chemical inertness of the surface. Consequently, adhesive interaction becomes weak and super low friction is achieved [22],... 

---