Research Article: Rotational switches in the two-dimensional fullerene quasicrystal

Date Published: January 01, 2019

Publisher: International Union of Crystallography

Author(s): M. Paßens, S. Karthäuser.


Local potential differences between the 36 and vertex configurations are identified within a two-dimensional dodecagonal fullerene monolayer. In a local area of the 8/3 approximant, rotational switching fullerenes on 36 vertex sites are revealed by scanning tunneling microscopy.

Partial Text

One of the key challenges in molecular electronics design is the stability of molecular states that are generally triggered from an ‘on state’ to an ‘off state’ or vice versa by application of an optical, magnetic or electric pulse (Zhang et al., 2015 ▸; Jeong et al., 2017 ▸). One way to achieve this goal is to use intrinsically bistable molecules which can be employed as switching units without major changes in their dimensions or electronic properties. Furthermore, they need to be adsorbed on a flat surface and wired to two metallic electrodes. Some prominent molecule families that are known to fulfill these conditions to a large extent are di­aryl­ethenes (Irie et al., 2014 ▸; Reecht et al., 2017 ▸), azo­benzenes (Alemani et al., 2006 ▸), and catenanes and rotaxanes (Dichtel et al., 2007 ▸; Fahrenbach et al., 2013 ▸). Their ability to switch in a controlled manner between two states by conformation change, ring opening/closing or redox reaction has been thoroughly studied in self-assembled monolayers or as single molecules using scanning tunneling microscopes, in break junctions, and in device geometries (van der Molen & Liljeroth, 2010 ▸; Feringa & Browne, 2011 ▸; Song et al., 2011 ▸; Jia et al., 2016 ▸). However, the problems with intrinsically bistable molecules are their often complicated chemical synthesis, the change in molecular properties caused by adsorption on a substrate, and an energetic disparity between the on and off states. A limited device performance, such as a degradation of the reliability or retention, often results.

A Pt3Ti(111) single crystal was purchased from MaTecK (Germany) and prepared by several ion sputtering (3 × 10−5 mbar Ne+ at 1 kV, 10 min) and annealing (1200 K, 25 min) cycles under UHV conditions. Under these conditions an overlayer consisting of two Pt layers is formed on the Pt3Ti(111) surface (Paßens et al., 2016 ▸). The cleanliness of the surface was checked by low-energy electron diffraction (LEED) and low-temperature ultra-high-vacuum (LT-UHV) STM.

As recently reported, the self-assembly of fullerenes on the 2Pt–Pt3Ti(111) alloy surface results in the formation of a two-dimensional dodecagonal QC (Paßens et al., 2017 ▸). The real-space LT-UHV-STM image of the quasicrystalline monolayer is depicted in Fig. 1 ▸. It is composed of triangles and squares and is superimposed with the tiling representation. The color coding of this square–triangle tiling indicates local areas of the 8/3 approximant, the approximant and the 36 approximant. The approximants are randomly distributed, and interpenetrating dodecagons and a small number of defects can also be detected. Consequently, the present dodecagonal QC corresponds to a random square–triangle tiling. Furthermore, the Γ1b phason strain, which describes the local deviation from ideal quasicrystalline order, was deduced from spot shifts identified in the fast Fourier transform (FFT) spectrum of the quasicrystalline fullerene monolayer (Paßens et al., 2017 ▸). Thus, fullerenes deposited on the 2Pt–Pt3Ti(111) single crystal form a two-dimensional dodecagonal QC with evident phason strain, while fullerenes normally form periodic structures on the (111) surfaces of metals including Pt.

To verify local potential differences between the fullerenes on the 36 and vertex sites we performed STS on each fullerene of a dodecagon. Differential conductance spectra taken over the molecules represent their local density of states and can thus be used to determine the respective highest occupied molecular orbital–lowest unoccupied molecular orbital (HOMO–LUMO) gap and the interfacial interaction. The HOMO–LUMO gap of single fullerenes in the gas phase amounts to 4.9 V. However, it decreases strongly to 3.7 V in the bulk crystal, to 2.7 V on metal surfaces like Au(111) (Paßens et al., 2015 ▸) and to only 2.5 V on 2Pt–Pt3Ti(111) (Fig. 4 ▸), indicating the strong interface interaction. Increased screening due to neighboring molecules and, more efficiently, due to interaction with metal surfaces leads to a reduction in the HOMO–LUMO gap (Torrente et al., 2008 ▸). Moreover, strong interfacial interactions between metal d states and the threefold-degenerate LUMOs of fullerenes may induce a lifting of the degeneracy and further broadening of the LUMO levels (Paßens et al., 2015 ▸).

The ultimate proof that fullerenes at the center of a hexagon exhibit increased mobility compared with fullerenes in the vertex configuration is their flipping ability. Fig. 5 ▸ shows the flipping of a fullerene in the 36 vertex configuration between two orientations, with an angle of 60° relative to each other. This rotational switching is induced by positioning the tip above the white-circled C60 in the left-hand STM image and applying a voltage pulse above a threshold value that induces rotation, i.e. 3.2 V for 30 s. A rapid change in tunneling current, in this case monitored after ∼7.5 s, indicates that a rotation of the C60 could be identified by scanning the same area again at a smaller sample bias voltage. The STM image in the center of Fig. 5 ▸ clearly shows that the white-circled C60 is rotated by 60° while the surrounding molecules have not moved. One reason for the drop in current is the position of the tip with respect to the orientation of the C60 cage. While the molecular orbitals rotate, the tip stays constant at the same position. At a positive bias voltage, higher tunneling currents should be measured if the tip is positioned above a pentagon, while lower tunneling currents should result if the tip is located above a hexagon. Due to the rotational switching of the central C60 by 60°, the position of the STM tip changes from above a pentagon to above a hexagon. By applying a second voltage pulse it could be shown that the rotational switching can be induced again. The tunneling current increases to its original value at the moment the molecule rotates. In the right-hand STM image of Fig. 5 ▸ the white-circled C60 again has the same orientation as in the left-hand STM image.

In order to gain further insight into the remarkable adsorption behavior of fullerenes on the 2Pt–Pt3Ti(111) surface, namely the formation of a quasicrystalline monolayer and the rotational switching of fullerenes on 36 local vertex sites, we have to consider the interfacial interactions in detail. The adsorption-energy differences in the 〈〉 directions, one of the nearest-neighbor directions in the dodecagonal monolayer, amount to 250 meV for different bridge sites (Paßens et al., 2017 ▸). In Fig. 8 ▸ only one of the possible rows in this direction is given. This is an interesting row for the possible adsorption on preferred bridge positions, since two distances between neighboring fullerenes, 0.96 and 1.10 nm, can be realized. However, neither distance is optimal with respect to van der Waals interactions between fullerenes (Fig. 8 ▸, lower part) (Gruznev et al., 2013 ▸). With an equilibrium intermolecular distance of 1.00 nm, a change to 1.10 nm would cost about 120 meV energy. Consequently, the fullerenes are tilted at a low energy cost in order to adopt the more preferred intermolecular distance of 1.04 nm as observed experimentally.




Leave a Reply

Your email address will not be published.