59th Annual Report on Research 2014

Spontaneous molecular self-assembly is a promising route for bottom-up manufacturing of two-dimensional (2D) nanostructures with specific topologies on atomically flat surfaces. Of particular interest is the possibility of selective lock-and-key interaction of guest molecules inside cavities formed by complex self-assembled two-dimensional host structures. Understanding of these systems could lead to the design of materials with highly specific selective adsorption, or catalysis of chemical compounds. Our studies focus on structures constructed with molecules of 1,3,5-tristyrylbenzene substituted by alkoxy peripheral chains containing n = 6, 8, 10, 12, or 14 carbon atoms (TSB3,5-Cn, see Figure 1) deposited on a highly ordered pyrolytic graphite (HOPG) surface.

Figure 1. Left: TSB3,5-Cn, where R represents the alkyl chain C_nH_2n+1. Right: TSB3,5-C6 optimized using Gaussian09 at the MP2/6-316(d,p) level. Carbon atoms are grey, hydrogen white and oxygen red. Longer alkane chains for n > 6 share the same basic structure.

Recent real-space imaging of reorientation of molecules adsorbed inside nanocavities formed in submonolayer films of such molecules show their potential as building blocks for self–assembled nanostructures [1-6]. Of particular interest are the possible "tunable cavities" observed in these systems [4], with opening diameters of 0.62, 1.04, 1.47, 1.89, and 2.31 nm for alkyl chains with n = 6, 8, 10, 12, and 14, respectively, see Figure 2.

Our approach focuses on understanding the mechanics of these structures from first principles. Initially we determined the optimal molecular configuration, electronic structure (including charge density), and intra-molecular potentials from quantum chemistry calculations. We utilized Gaussian09 [7] and performed calculations sequentially at the Hartree-Fock (HF) and 2^nd order Moller-Plesset perturbation theory (MP2). The basis set was incremented from 321g to the Pople split-valance double-zeta 6-31g(d,p) basis set [8]. The calculations result in bond lengths, bond angles, dihedral angles and Mulliken charges for each atoms. Our results are comparable (differences = 1-5%) with tabulated CHARMM force field parameters (CGenFF, [9]) and facilitated the identification of distinct atom types (e.g., OG301, HGR61, CG2R6, CG2DC1, etc.). Currently we use the "elastic constants" and non-bonded Lennard-Jones parameters from Ref. [9], but ab initio calculations based on our molecules are currently under way for added consistency. Optimized structurues for small clusters of TSB3,5-C6 molecules are in qualitative agreement with observations (Fig. 2).

Figure 2. Left panel: expected self-assembled structure for TSB3,5-C10. Right panel: potential energy minimization of a small cluster of TSB3,5-C6 molecules using a combination of our MP2/6-31g(d,p) ab initio calculations and CGenFF CHARMM parameters [9].

Figure 3. Molecular dynamics simulation of 32 TSB3,5-C6 molecules at 100 K. We utilize periodic boundary conditions with a commensurate 4 x 4 TSB3,5-C6 superlattice at 11.7¡ orientation relative to the graphene lattice. For reference, the computational cell is ~ 12.5 nm, and the graphene cell is 0.246 nm.

A triangular super-lattice of TSB3,5-C6 molecules (4 x 4 cluster, ~ 12.5 nm sides, commensurate, at an angle of 11.7¡ to the graphene 0.246 nm lattice vector) was simulated via molecular dynamics with periodic boundary conditions (computational box shown in the xy plane in Figure 3). The A six-layered A-B sequence of rigid graphene planes was covered by 32 TSB3,5-C6 molecules. Figure 3 shows the results of the simulations at 100 K. Clearly, inter–digitation of the alkane chains has an important effect on the adlayer, and the general morphology is similar to what is observed experimentally [4]. We are currently varying the super-lattice parameters to study if there are any surface tension effects, and performing simulations at higher temperatures to observe structural transitions and perform simulated annealing. Computational work (ab initio and molecular dynamics) for longer alkyl chains (n > 8) is planned for 2014-2015.

REFERENCES

1. F Mathevet et al., Adv. Mater. 18, 2954–2957 (2006).

2. G Schull et al., Nano Lett., Vol. 6, No. 7, 2006.

3. G Schull, et al., J. Phys. Chem. C 112, 14058–14063 (2008).

4. C Arrigoni, et al., J. Phys. Chem. Lett. 1, 190–194 (2010).

5. J.V. Barth, Annu. Rev. Phys. Chem. 58, 375–407 (2007).

6. A.J. Groszek, A. J. Proc. R. Soc. London, Ser. A 314, 473 (1970).

7. J.A. Montgomery Jr. et al., Gaussian 09, rev a.1; Gaussian Inc., Wallingford, CT, 2004.

8. M. Head-Gordon, J.A Pople, and M.J Frisch, Chem. Phys. Lett 153, 503–506 (1988).

9. B. R. Brooks et al., J. Comp. Chem. 30, 1545-1615 (2009).