Scanning Tunneling Microscopy of molecules on metals

Abstract

Scanning Tunneling Microscopy (STM), a relatively new technique capable of observing atom-sized features, has recently made significant inroads into observations of molecules on surfaces. At first thought it might seem that a microscopy which can image atoms should be trivially capable of observing molecules. There are several reasons why this is not the case. First, because the atoms within a molecule are already bonded to one another, molecules are frequently only weakly affixed to surfaces and other molecules, and consequently may be mobile at room temperature. A second difficulty arises because most molecules are electrically insulating. Since STM detects atoms by sensing a small flow of current between an atomically sharp probe and the sample, a perfectly insulating molecule might be expected to be invisible with the STM. Luckily, it is often found that chemical interactions between molecules and a metal surface are sufficiently strong to change the electrical properties enough so that thin layers of insulating molecules are easily detected. Another significant problem involves the possibility that the tunneling tip is brought so close to the conducting support surface that molecules are essentially squeezed out of the vacuum gap and remain undetected. These difficulties have caused a significant delay in the success of applications of STM to high resolution molecular imaging, as opposed to early STM successes with metal and semiconductor surfaces. Our data indicates that each of these problems can be important, and that steps can be taken to reduce or eliminate them. We illustrate the recent success of STM at this new task with the imaging of a few organic molecules on metal surfaces. Our first studies of molecules involved ordered coadsorbed CO and benzene on Rh(111) surfaces. This system shows a 3x3 structure in low energy electron diffraction (LEED) for a 2/1 CO/benzene ratio (1) and a c(2√3 x 4)rect structure for a 1/1 ratio (2). For the 3x3 structure, it was found that benzene was easily imaged while the CO image contrast was substantially smaller (3). In addition to imaging well-ordered terraces, we were able to observe benzene binding near step edges, diffusion (possibly induced by the STM) of benzene at defects, and internal structure within the benzene molecules. The sharpest images possessed only 3-fold symmetry, presumably because 3 of the 6 C-C bonds are above substrate metal atoms. Further reductions in the image symmetry associated with asymmetric tips were frequently observed. Subsequent observations of the c(2 √3 x4)rect structure revealed the three domains present for this structure and frequent domain changes at step edges. For this system we also found small protrusions to occupy the sites assigned to CO in the LEED model (4) so that this molecule can also be imaged. Images of disordered samples were difficult to interpret because of possible diffusion during the image acquisition. In subsequent measurements we studied Cu-phthalocyanine, which is of interest due to the catalytic activities of its central metal atom, on different substrates in order to better understand and exploit the consequences of variations in the molecule-surface interaction. In preliminary work, using submonolayer coverages on Si(111) or Au(111), we were not successful in obtaining good images which showed the 4-fold symmetry of this molecule. On Cu(100), however, atomic resolution images of this molecule were readily obtained for coverages up to 1 monolayer (5). At low coverages the molecules occupy two binding sites related by a mirror plane and show little tendency to form islands, although there may be some local orientational order. Molecules near step edges appeared to overlap the steps, with two lobes on the upper terrace and two lobes on the lower terrace. It was possible to adjust the operating conditions so that tip-molecule interactions appear to induce translational motion of the molecules. The lattice of the Cu(100) substrate was also observed but the molecules appear less stable, or simply absent, under conditions where the metal corrugation is large. Packed ordered arrays of molecules were observed near 1 monolayer coverage with domain structures in agreement with earlier LEED observations (6). When the coverage was increased slightly above 1 monolayer we found that we were unable to attain high resolution images of second layer molecules. At coverages slightly above 1 monolayer, we observed protrusions above a packed, ordered first layer, which appear diffuse and unstable. The diffuse character of the second layer molecule images strongly resembles that of images of sub-monolayers on Au(111) and Si(111). We suspect that the poor resolution results from inadequate binding of the molecules to the underlying layer. This does not imply that the molecules are too poorly conductive to be imaged, because the apparent height of second layer molecules is quite similar to that of first layer molecules. But they may be moved about by thermal effects or tip-molecule interactions. Although it is still in its infancy, these results suggest that STM will become increasingly important for studies of surface chemistry. Improvements in the apparatus aimed at increasing the image acquisition rate, allowing operation at different temperatures, and employing more elaborate sample and tip preparation techniques should extend the impact of this technique to a broader range of systems. A better understanding of the origin of the image contrast is certain to result both from empirical results and more quantitative theoretical analyses. Recent work will hopefully illustrate these points. It is our hope and expectation that STM images of reactive molecules will help to provide insight into the atomic scale details of surface mediated chemical processes.