ELSEVIER Synthetic Metals 84 (1997) 317-318 Ab-initio calculations of the gold-sulfur interaction for alkanethiol monolayers K.M. Beardmore, J.D. Kress, A.R. Bishop, and N. Gronbech-Jensen Theoretical Division, MS-B258 Los Alamos National Laboratory, Los Alamos, NM 87545, USA Abstract We have performed ab-initio geometry optimization of SCH, absorbed on cluster models of Au( 11 l), to examine the gold-sulfur- carbon head group energetics for self-assembled monolayers of alkanethiols on gold. The ab-initio calculations are performed using gradient-corrected density functional theory. The data obtained from these calculations, are used to construct a realistic empirical potential function to describe the complete headgroupdu(ll1) energy surface. In particular, the dependence of the energy of the system on the Au surface-sulfur separation, and the surface-sulfur-carbon angle is determined for positions representative of the entire gold( 111) surface plane. Keywords: Ab-initio quantum chemical methods and calculations; Density functional calculations; Molecular dynamics, lattice dynamics; Computer simulations; Models of surface and interface chemistry and physics; Self-assembly using surface chemistry. I. Introduction Self-assembled monolayers (SAMs) are a class of molecular assemblies that are prepared by spontaneous adsorption of molecules from solution onto a solid substrate. Many self- assembly systems have been investigated, but monolayers of alkanethiolates on gold are probably the most studied SAMs to date [I ,2]. Most theoretical studies of alkanethiols on gold(ll1) surfaces assume the headgroup positions are restricted to the vicinity of the 3-fold hollow sites in a hexagonal packing [2], but recent experimental and simulation data have questioned this [3-51. Therefore, we have begun to construct a model of the S-Au interaction over the entire Au(ll1) surface, that incorporates realistic energies, energy barriers, and force constants, for use in molecular dynamics (MD) simulation. 2. Ab-initio calculations Electronic structure calculations were performed using gradient-corrected density functional theory (GC-DFT) with the program Gaussian94 [6]. The Becke exchange and the Lee- Yang-Parr correlation functionals were used. Accurate all- electron double-zeta valence plus polarization basis sets (6- 3lG*) were used for S, C, and H. The LANLlMB effective core potential [7] was used for Au, where only the valence shell electrons (5d, 6s) are represented explicitly with a minimum basis set. The Au(lll) surface was assumed to have the (1x1) structure, consistent with experimental observations [4,8] that the ( &x23) reconstruction of clean gold disappears upon monolayer formation, and was represented as a cluster. The largest cluster studied consisted of 17 Au atoms (10 surface and 7 second layer atoms). The energy of Au,,SCH, was calculated for the S atom above the atop, bridge, 3-fold fee and 3-fold hcp sites as well as 5 other unique sites midway between the first four sites. The SCH, geometry was fixed close to that found in 0379-6779/97/%17.00 0 1997 Elsevier Science S.A. All rights reserved PII s0379-5779(9G)04019-2 HSCH,, the angle between the Au surface normal and the S-C bond (denoted by QAUSc) was fixed at 180°, and the Au surface to S distance ( ZAUs) was optimized. Then further calculations were started for smaller clusters, where BAuSC as well as ZAiis was optimized. 3. Results and discussions Our preliminary results, and those from a similar calculation by Sellers et al. [9] are given in Table 1. From our calculations, the hcp site is the stable minimum and the bridge site is the barrier between 3-fold sites, although the atop site is only 0.5 kcal/mol higher that the bridge site. There are two points of disagreement with the results of Sellers et al. [9]. Our calculated Au-S distances are around 0.4 A larger, and the optimum Sangles vary considerably from 104”. For either 3-fold site, we do not obtain a stable minimum at the angle of 180”, as reported in Ref. [9]. However, the force constants reported for that configuration are relatively small, suggesting that it is only marginally stable. 4. Empirical potential fitting To obtain a continuous potential, a three dimensional function was fit using the Q,,,=lSO” data. The data for the nine points was first least squares fit using a Morse function parameterized as v(z) = De@%(z-R.) _ 2De,e-p4z-Re) We prefer to use a this form, rather than, say a quadratic harmonic function as it is asymmetric and remains finite as z AuS increases. Thus it is expected to better extrapolate energies for larger Au-S distances, if this is necessary. For the fitting, the relative energies of the minima, were shifted, so that the deepest (hcp site) had a value of -28 kcal/mol. The parameter, p,, showed little variation over the nine data sets, so