Cross-Linked Gold Nanoparticle Assemblies: What can we Learn from Single Flat Interfaces?

Cross-linked gold nanoparticle (GNP) assemblies are valuable for a variety of applications, such as transducers for strain or gas sensors. To pave the way for understanding their sensing behavior on an atomistic scale, we ask whether their properties can be modeled by a single, flat, particle–particle interface. Employing reactive force field (ReaxFF) molecular dynamics simulations and a tight-binding density functional the- ory approach to coherent tunneling, we find that for alkane dithiolates, where most molecules will typically bridge between the gold surfaces, the interparticle distances, as well as the conductivity of the assembly, can be modeled to even quantitative accuracy with a single-interface model — if comparing to sufficiently large GPNs for which the flat surface is a good approximation. For alkane monothiols, where each molecule is only attached on one side, the difficulty of estimating surface density and the resulting degree of interlacing results, in our case, in underestimating the interparticle distances by around 5 Å. The increase of these distances (and the decrease of conductivity) as going to longer alkane chains, however, is still well reproduced. We discuss shortcomings of ReaxFF, such as not being able to describe thiol physisorption and producing spurious reactions between dithiolates, as well as factors influencing the validity of a single, flat, particle–particle interface model. Our model reduces the system complexity significantly compared to simulating entire GNP assemblies, enabling further mechanistic investigations of sensing properties of these systems with atomistic approaches, and, potentially, screening of ligands for specific sensing applications.