Dual Modulation Scanning Tunneling Microscopy: a Quest for Subnanometer Chemical Contrast on Thin Films

The unfathomably high density of transistors on modern day computer chips has given rise to generation defining technologies, such as smartphones, robotics, and artificial intelligence. The immense improvements in technology made over the past decades, such as the downscaling of the transistor, were enabled by the ever-increasing control that humankind has over matter. Nowadays, in many material applications we are pushing the limits of what is possible by thinning layers down to mere nanometers or even smaller. For such applications, only a few misplaced atoms could have significant consequences and so a technique that allows us to distinguish chemical components with atomic resolution could prove vital for material engineering in the near future.

Scanning tunneling microscopy is a technique that is famous for its ability to capture atomic resolution images of materials. With this technique, an atomically sharp conductive tip is brought to within a nanometer of a (semi-)conductive surface. By applying a voltage between the surface and the tip, electrons in the sample and tip can quantum mechanically tunnel between them. The tip is then controlled with piezos: both its lateral position as well as the height with respect to the surface. By scanning a raster of points, the scanning tunneling microscope generates atomic resolution images in which both topographic and electronic information are intertwined.

The electronic information obtained through scanning tunneling microscopy can serve as a chemical fingerprint. Fortunately, there exist techniques that can (partially) untangle the topographic and electronic information. Most notably, dI/dV and dI/dz spectroscopy, which measure the differential conductivity and local work function, respectively. The first provides information on the availability of electronic states at particular energies, which can originate from the localization of electronic states around particular atoms. The latter provides information on surface dipoles, which can originate from chemical bonds.

In this thesis, we set out to investigate how these techniques could be combined to measure conventional scanning tunneling microscopy images together with maps of the differential conductivity and local work function. The resulting methodology we developed is dual modulation scanning tunneling microscopy, which therefore also forms the main title of this thesis. Chapters 2 & 3 focus on the development of dual modulation scanning tunneling microscopy. The Chapters that follow, cover the various model systems we studied. During the investigation, each system revealed their own interesting particularities with respect to surface physics.

The decanethiol molecular self-assembled monolayers on Au(111) we studied created a nanoscale pattern in the local work function based on how the molecules were lying on the surface. In molecular monolayers the local work function is often lowered by the electrical dipole in the chemical bond between the molecule and the surface or the electrical dipole over the molecule itself. However, the first is negligible for decanethiol molecules chemisorbed onto Au(111) and the latter is only present when the molecules are standing upright. However, we found using dI/dz spectroscopy that lying-down decanethiol molecules also lowered the local work function with their physisorbed hydrocarbon tails. We attribute this effect to Pauli repulsion between electrons on the molecular tails and electrons in the substrate: the so-called ’pillow’ effect. We also investigated these monolayers after oxidation and found that this local work function variation had disappeared, which demonstrates that the heads of the molecules were no longer chemisorbed to the surface. These findings form Chapter 4 of this thesis.

The main model system we studied, was vanadium on Si(111). By depositing and subsequently annealing varying amounts of vanadium, we grew and investigated vanadium disilicide clusters on Si(111); a material system that has only had limited attention in the field of surface science. When depositing a large amount of vanadium (a 30 nm thick layer), we found three main cluster types: flat top clusters, faceted clusters and hut clusters. However, at intermediate initial surface coverages (8 monolayers) the clusters instead form elongated structures with markedly high aspect ratios. Examining the shape of the clusters, we determined that this transition stems from a competition between edge formation energy and strain relaxation, as described in the model by Tromp and Tersoff.

While lowering the initial surface coverage further (4 monolayers), we found that this material system can also grow nanoribbons. Furthermore, we found evidence of resolidified droplets that seem to have moved along the temperature gradient that was present on a sample. We propose that, given the presence of a eutectic point at the silicon-rich side of the vanadium-silicon phase diagram, the latter observation can be explained by the formation and movement of eutectic droplets. Upon cooling down of such a eutectic, vanadium disilicide clusters would be formed with the excess silicon being push out to the surface of the clusters: a phenomenon known as spinodal decomposition. In this light, a reconstruction we had observed on a flat top cluster could be explained: as it did not match well with vanadium disilicide but did match exceedingly well with the (√3×√3)R30° reconstruction of silicene, the 2D graphene analogue of silicon. Finally, using the combination of dI/dV and dI/dz spectroscopy on the nanoribbons revealed a Dirac-like electronic structure, which can be fitted with theoretical models to find a Fermi velocity that matches well with that of silicene. This lead us to the conclusion that vanadium on Si(111) is a material system that is capable of growing the Dirac material silicene through spinodal decomposition. Chapters 5 and 6 cover our investigation of the vanadium silicon system.

The final model system we studied were vanadium/silicon and molybdenum/silicon multilayers. Such multilayer structures are essential components in extreme ultraviolet optics. We provided a proof-of-concept for preparing these structures for cross-sectional scanning with scanning tunneling microscopy. First, we etched notches along crystallographic planes in silicon wafers with tetramethylammonium hydroxide. Then we grew the multilayers on the unaffected side of the silicon wafers, after which the wafers were divided into samples. Then we broke the samples by fixing one end and applying pressure to the other end. In doing so, the samples broke each time from the notches along crystallographic planes, resulting in mirrorlike cleave surfaces which are sufficiently smooth for scanning tunneling microscopy. Moreover, using Kelvin probe force microscopy, we found that the layers of the multilayer structures could be distinguished by their local work function, demonstrating that dI/dz spectroscopy could provide information about the structural properties of the cross-section of multilayer structures, e.g., hopefully about the mixing of the layers.

In conclusion, we developed a scanning tunneling microscopy technique that provides two spectroscopic signals and a topographic signal simultaneously. We then used scanning tunneling spectroscopy, with support from other surface science techniques, to explore various thin film model systems. We found that the spectroscopic signals measured with dual modulation scanning tunneling microscopy can provide useful insight into a myriad of surface properties. However, whether or not measuring the dI/dV and dI/dz spectroscopy simultaneously provides a clear step forward towards distinguishing chemical contrast on the atomic scale, we have not been able to show conclusively.