Reactions of Lithium with Small Graphene Fragments: Semi-Empirical Quantum Chemical Calculations
- PDF / 417,737 Bytes
- 6 Pages / 414.72 x 648 pts Page_size
- 32 Downloads / 130 Views
random stacking of independent graphene fragments which thus can bind "interstitial" Li on both sides and hence twice the capacity of graphite. Other models invoke Li 2 "molecules" which are epitaxially formed over first-neighbor hexagons [6] or involve multiple monolayers of Li over the same graphitic fragment [7]. Alternatively, we have proposed [1] that residual hydrogen enhances the lithium affinity; here we extend that idea to show that fragment edges, whether protonated or not, provide for "excess" Li capacity. Most recently, a group from MIT and Tokyo has reported simulations which closely resemble ours in character but with slightly different results [8]. In this work, we study the Li interaction with the disordered carbohydrogen alloys modeled as aromatic rings with some unsaturated edge valencies. All the simulations were performed at the semiempirical MNDO/AM1 level, the reliability of which was checked in earlier work by comparison with ab initio results on small PAH'es [1]. Computational Methods Over the years, several semiempirical methods have been developed to speed up and simplify calculations for larger molecules. Such approximations most often consider only the valence electrons, while parametrizing the ionic cores. Further improvements such as zero differential overlap (ZDO) and modified neglect of differential overlap (MNDO) [9] are included as well. AMI is just a newer version of MNDO, in which the core-core repulsion functions are modified and fitting functions for two center integrals are reparametrized [10]. AMi has proven to be reliable for aromatic carbon systems, however Li has not been parametrized within this system. Thus, we use the older MNDO set of parameters for Li only, a compromise which has produced relatively good qualitative results in related systems [11]. All of these semiempirical calculations were carried out with the molecular orbital software MOPAC [12]. In order to check reliability of our AM1 calculations, we compared some of our results to ab initio values for the same geometries obtained from the Gaussian-94 program [13]. As discussed elsewhere [1], our methods suffer from overestimating steric repulsions, due to the small basis set. Also, the potential energy of CLi 2 group is not handled properly, but this should not be a problem since our calculations will not encounter such local insulated geometries. Finally, it is also worth noting that the systems under study exhibit many local minima which makes the search for the global ground state extremely time consuming and difficult. Instead, we start with many initial configurations and draw our conclusions independent thereof. Results For completeness, we first summarize results of our simulations on hydrogen-terminated graphite clusters published elsewhere [1]. The smallest interesting test molecule is the 4-ring hydrocarbon C 16H1 0 (pyrene) because it contains two distinct types of hexagons terminated by either two or three hydrogens. We shall refer to these as "2H" and "3H" hexagons, though one can apply the termi
Data Loading...
