Towards “Mechanochemistry”: Mechanically Induced Isomerizations of Thiolate–Gold Clusters

Abstract

Stretching the imagination: Based on simulations it is proposed that atom-manipulation techniques, such as atomic force microscopy (AFM), could be used to draw and manipulate monoatomic nanowires using thiolates anchored to gold clusters on surfaces (see picture, gold Au, yellow S). At the same time, the force-versus-extension curve is diagnostic for the isomerizations of the underlying covalently bound species. The investigation of the electrical, mechanical, and chemical properties of atomic gold wires and the nanopoint contacts connecting two metallic surfaces with a view to developing molecular electronic devices based on nanoscale metallic wires is an increasingly popular field of research.1–6 An intriguing possibility is to use mechanical energy as a tool to construct such nanoscale architectures. For example, suspended monoatomic gold nanowires7–9 have been detected by using mechanically controllable break junctions or atomic force microscopy. Recent theoretical work has focused on the modulation of mechanical and electrical conduction properties of such nanowires bridging two gold surfaces by the inclusion of defects, such as sulfur atoms and thiolate molecules.10, 11 However, the role of these defects in the formation and growth, as well as their implications for the isomerization chemistry, of these wires remains an open issue. We recently reported ab initio molecular-dynamics simulations that, surprisingly, also lead to such a nanowire.12 In this simulation a single chemisorbed thiolate molecule was pulled from a gold surface, a situation akin to corresponding experiments.13 Thus, these simulations suggest that a single thiolate molecule may be used as a moleculer “hook” which allows the controllable extraction of monoatomic gold nanowires from bulk metallic surfaces. However, for this effect to be of practical importance it must be transferable to other nanoscale materials, such as gold nanoparticles deposited or grown on semiconducting substrates.2, 14 Relevant to this discussion is the observation12 that the pulling process resulted first in the temporary formation of a “localized gold cluster” which was initially pulled from the surface along with the thiolate. This situation suggests that the underlying mechanism for drawing monoatomic gold wires may be the reorganization of the metal–metal bonds within the gold-cluster “intermediates” and would suggest that the process is transferable to interfaces other then those involving flat surfaces. In principle, this could open up possibilities for “mechanochemical wiring” of hybrid molecular–metallic nanostructures at will. Is it easy to estimate if the above-mentioned processes might be possible on small clusters instead of surfaces? The bonding of thiolates to gold clusters offers a wide variety of structures including some in which there are fewer, but stronger, AuS bonds than found with bulk surfaces.15, 16 In addition to the coordination-number issue there are also energetic considerations which are more difficult to judge. Typical thiolate–gold-cluster binding energies (from gradient corrected density functional calculations) are in the range 2.5–3 eV,15 which is between the AuAu bond energy in Au2 (about 2.3 eV) and the cohesive energy of the face-centered-cubic (f.c.c.) solid (approximately 3.8 eV). Furthermore, the chemisorption energy on bulk terminated surfaces is of the order of 1.5–1.8 eV16 depending on the method used and the binding site. Still, it is possible to draw nanowires from surfaces because the decrease in coordination number occurs in conjunction with dynamic restructuring.12 Thus, simple estimates do not lead to a faithful assessment, so we performed electronic-structure calculations based upon Kohn–Sham density functional theory.17 To mimic a typical mechanical pulling experiment involving a molecule attached to a supported cluster we have used Au3SCH3 and Au5SCH3. Such small species represent limiting cases of molecular bonding in which the few AuS and AuAu bonds present are strong. Thus, these low-coordination cases are very different from the situation encountered at extended surfaces. The following constrained geometry relaxation model was applied (Figure 1, see refs. 18, 19 for related approaches): one gold atom was constrained to move in a plane P1 whereas the methyl carbon atom was confined to another parallel plane P2. Starting from the fully optimized structure of the cluster–thiolate adduct, zi=0 Å, the distance between the planes z was increased stepwise in increments of Δz=0.05 Å. At each step zn the entire system was reoptimized ensuring that the two atoms were confined to their respective planes. The extension force, which would correspond to the experimentally measured pulling force, is obtained from the slope of the total energy versus the extension (zn): F=−dE/dz≈−ΔEn/Δzn. Finally, the calculation for the smaller adduct was checked using both spin-polarized and unpolarized calculations, which gave identical results. Initial (top) and final (bottom) configurations of the rupture calculations of Au5SCH3. The planes P2 and P1 constrain the carbon atom and the most distant gold atom, respectively, and zi is the initial and zf the final distance between these planes. Both cluster sizes yielded qualitatively similar findings: transformation of the compact gold cluster to an open monoatomic wire followed by rupture of a gold–gold bond within the wire. Hence, we can focus our discussion on the larger, and more complex, Au5SCH3 system (Figure 2). The calculation is initiated with the lowest-energy isomer,15 in which the Au atoms assume a planar closest packing structure (A-I) and the thiolate is bound symmetrically to two gold atoms. Upon pulling, this twofold coordination of the sulfur atom remains intact initially, whereas the five gold atoms of the cluster itself gradually adopt a circular structure (A-II). As shown by a clear change in the slope of F(z) at z≈1.1 Å (Figure 3), there is a smooth transformation from a closest-packed triangular structure to the open Au5 ring, A-II. This family of structures A is stable upon elongation of up to about 3.8 Å where there is a discontinuous rearrangement with the five-membered ring collapsing into a four-membered ring with a capping atom, (structure B), which induces a jump in the force-versus-extension curve. Thus, both these gradual and sudden changes in the structure of the cluster can be discerned by distinct features in the force profile (Figure 3). After another isomerization of the cluster at z≈4.3 Å leads to structure C, the coordination number of the sulfur atom reduces from two to one at z≈4.8 Å (structure D-I). The transformation from D-I to D-II can also be detected in the F(z) profile. Most importantly, the next transition is to a wirelike structure E, which is similar to the species reported in our surface simulations.12 Finally, wire rupture occurs leaving two fragments, a gold dimer (Au2) and a bent Au3 molecule which is anchored to the thiolate (Au3SCH3; structure F). Importantly, this entire sequence of “isomerizations” can be readily detected in the force profile. Side view (left) and top view (right) of the configuration of the Au5SCH3 system at the elongations as marked in Figure 3. The bold gray lines are parallel to the constraint force. Evolution of the total energy (bottom) of Au5SCH3 relative to the fully relaxed initial configuration A-I and the corresponding extension force F (top) as a function of z. Views of the labeled structures are displayed in Figure 2. Estimated computational errors in F are of the order of 0.1 nN; note that the sign convention of the force is different from the one used in ref. 12. Overall, the force curve of this transformation process (Figure 3) is qualitatively and even quantitatively similar to those that are typically observed in rupture experiments using pure gold–gold nanojunctions.9 Specifically, our cluster calculations reproduce the general saw-tooth behavior, the vanishing force F≈0 nN when wire formation begins, a maximum force of around 2 nN before breaking, and a smaller rupture force of about 1.5 nN. Our calculations also underscore an important observation: it is possible to employ external mechanical forces to create high-energy chemical species, as demonstrated by the formation of stable configurations (such as, D-I and E-I, which are 1.36 eV and 2.25 eV above the ground state A-I, respectively) where the total force is zero. In addition, the energetics and corresponding structures generated in the rupture process give some unique insights into the complex chemistry occurring during the drawing. The rich isomerization chemistry of gold clusters15 allows the system to adopt substantially different configurations during the elongation process as a result of rearrangements in the chemical bonding pattern. More precisely, the rupture of the smaller adduct, Au3SCH3, results in the final fragments, Au2 and AuSCH3, in their ground state with ΔErupture(Au3SCH3) ≈ Emin(Au2) + Emin(AuSCH3)−Emin(Au3SCH3) =1.67 eV. Similarly, for Au5SCH3, ΔErupture(Au5SCH3)=3.38 eV is in excellent agreement with the value of Emin(Au2) + Emin(Au3SCH3) + ΔE*−Emin(Au5SCH3) =3.41 eV, where, ΔE*=0.48 eV is the energy difference between Au3SCH3 with a bent Au3 unit and the lower-energy triangular Au3 isomer.15 Thus, the overall picture that emerges is that the mechanical pulling force literally induces a series of isomerization reactions to stable intermediates which ultimately lead to energetically favorable fragmentation of the cluster itself. In conclusion, by examining small thiolate–gold cluster adducts we have demonstrated that under an applied pulling force the cluster unfolds from a compact form to an extended wirelike configuration. Supplemented by earlier findings of an analogous nanowire formation from bulk metal surfaces, this result implies that monoatomic wires may be drawn from a wide variety of gold-based nanoscale architectures including molecular and nanoscale clusters as well as macroscopic leads. Taken together, this opens up possibilities to use gold clusters, firmly attached to an insulating or semiconducting support, to construct nano-electronic assemblies by atom-manipulation techniques. Most interestingly, wire formation in the sense of a sequence of chemical transformations is found to proceed along a well-defined reaction coordinate, which is externally imposed by the applied force. Along this preselected path a series of mechanically induced isomerization reactions occurs, which can be monitored by the behavior of the force–elongation curve. These observations could serve as a general method for future investigations where mechanical rather than thermal, light, or electrical energy is used to induce controlled chemical reactions involving covalent bonds.