Migration of gold atoms into a thiol-bonded molecular self-assembled monolayer, forming a cluster exhibiting a Coulomb staircase

TL;DR

This work reveals that thiol-terminated SAMs on gold can host spontaneous, nanometer-scale gold clusters that give rise to Coulomb-blockade and staircases in single-electron transport when integrated into graphene–Au junctions measured at cryogenic temperatures. Through a combination of transport experiments, AFM/UFM imaging, XPS, and DFT modeling, the authors show that gold atoms migrate into the SAM under thiol anchors to form metallic islands, whereas amine anchors trap Au and suppress cluster formation, offering more reproducible device behavior. The experimental Coulomb staircases are well described by a classical Coulomb-blockade model augmented for graphene electrodes and background charge, yielding dot sizes of a few nanometers and charging energies consistent with nm-scale clusters; occasional nanoparticle insertion further validates the cluster hypothesis. Additionally, the study demonstrates tunable background-charge gating and, in some devices, memristive switching and negative differential resistance arising from multi-dot transport pathways, suggesting practical routes for multifunctional nanoelectronic components and highlighting amine anchors as a more reliable alternative for scalable SAM-based devices. Overall, the work provides a framework for controlling nanoparticle formation in molecular junctions and points to design principles for robust, low-power molecular electronics using anchor chemistry and vertical graphene contacts.

Abstract

Thiol-based self-assembled monolayers (SAMs) on gold surfaces are one of the fundamental building blocks of molecular electronics. The strong chemical affinity of the gold and sulfur (Au-S) enables the formation of close-packed SAMs, but it also has recently been found to create a dynamic interface where surface reconstruction can occur under illumination, even with ambient light. This reconstruction may facilitate migration of gold atoms, potentially leading to in-situ formation of gold clusters. However, research on this mechanism often centers on Au(111) crystalline surfaces and flicker-noise measurements. Electron transport in ensembles of molecules in lithographically defined junctions has remained largely unexplored at cryogenic temperatures. In this study, we observe single-electron phenomena characterized by reproducible Coulomb staircases across various long-chain alkanethiol SAMs, which fit the Coulomb-blockade theory of nm-sized metallic nanoparticles. We find no such current steps in samples with amine, rather than thiol, anchors. Additionally, we find that by adding a bipyridyl functional group, these phenomena can be harnessed for memristive switching and negative differential resistance. These findings indicate that the generally observed lack of reliability and reproducibility of molecular devices may be alleviated by using amine anchors instead of thiols to avoid nanoparticle effects. Conversely, the spontaneous formation of the nanoparticles could potentially be controlled and used to achieve useful functionalities, offering new pathways for designing multifunctional nanoelectronic components.

Figures (14)

• Figure 1: (a) Cross-sectional schematic renderings of the junction, showing a monolayer of molecules assembled on a gold layer inside a hole etched through an insulating layer. The top contact is graphene. A gold cluster, here shown as a disk beneath the graphene, may form in a pristine monolayer as Au atoms migrate from the surface, as indicated by the arrows between diagrams. (b)-(g) A collection of junctions with various areas from different batches but showing similar pronounced staircases in the $I$--$V$ characteristics (the first and third rows) with corresponding multiple peaks in the conductance ($G$) spectra (the second and fourth rows) at 4.2 K. Curves with different colors represent consecutive sweeps, showing reproducibility. (h) Height map captured with PeakForce tapping AFM for a microwell containing a thiol-anchored SAM, showing only a few white specks that may be clusters. (i) High-resolution AFM height map of the red box in (h). (j) 3D view of (i), showing a single bump under the graphene. (k) Image of a bump in another microwell junction using ultrasonic force microscopy, showing a dark ring of unsupported graphene and a firm region in the middle of the bump (lighter), which we attribute to a gold cluster less than 15 nm in diameter and 10 nm in height, possibly a little larger than one that gives rise to a Coulomb staircase.
• Figure 1: Raman Spectroscopy of a representative piece of graphene on a device.
• Figure 2: $I$--$V$ and conductance (G) measurements (blue dots) and their corresponding fits (orange curves, see text), for a representative variety of junctions. (a), (b), (g) have a C16S SAM, (c), (h), (i) have HSC11BIPY with C10S2 supporting molecules, (d) has C18S, and (e) and (f) have HSC11BIPY with C8S2 supporting molecules, where 2.2 nm and 3 nm gold nanoparticles were deliberately introduced, respectively. Except for (e) and (f), for which the measurements were conducted at 77 K, the rest of the measurements were carried out at 4.2 K.
• Figure 2: Comparison of the I--V and G--V curves for four example junctions displaying Coulomb staircase at 77 K (blue) and 300 K (red). The data at 300 K were taken before low-temperature measurements. These are all HSC11BIPY junctions.
• Figure 3: Temperature-dependence measurements of Junction 1. (a) Experimental data, (b) fits to each of the curves in (a) The color of each curve corresponds to the temperature given by the colorbar to the right of each plot, which includes any other forms of energy broadening. A sudden change occurred in the junction before the fifth trace from the top while warming from low temperature (at around 70 K). The corresponding main change in the fitting parameters (from the set labeled 1$^\prime$ to that labeled 1$^{\prime \prime}$ in Table \ref{['tab:parameters']}) is in the background charge $Q_0$, where the values of $Q_0/e$ below and above 70 K are $0.12\pm0.01$ and $-0.26\pm0.01$, respectively. The other fitting parameters change by less than 20%.