An electrically actuated molecular toggle switch

Molecular electronics is considered a promising approach for future nanoelectronic devices. In order that molecular junctions can be used as electrical switches or even memory devices, they need to be actuated between two distinct conductance states in a controlled and reproducible manner by external stimuli. Here we present a tripodal platform with a cantilever arm and a nitrile group at its end that is lifted from the surface. The formation of a coordinative bond between the nitrile nitrogen and the gold tip of a scanning tunnelling microscope can be controlled by both electrical and mechanical means, and leads to a hysteretic switching of the conductance of the junction by more than two orders of magnitude. This toggle switch can be actuated with high reproducibility so that the forces involved in the mechanical deformation of the molecular cantilever can be determined precisely with scanning tunnelling microscopy.

(56 mg, 67 µl, 420 µmol) were added under argon and the tube was quickly capped. The reaction mixture was heated at 80 • C for 15 h. The completion of the reaction was checked by TLC (hexane:EtOAc = 40:1). After cooling, the reaction mixture was diluted with diethyl ether (20 ml), filtered through a pad of silica gel (20 g, diethyl ether), and the filtrate was concentrated in vacuo. The crude product was purified by column chromatography on silica gel (300 g, hexane:EtOAc = 40:1) to provide the title compound 2 (226 mg) as a colorless oil in 75% yield (R f = 0.29, hexane:EtOAc = 40:1). 1   ),    Supplementary Fig. 8, and the angle of the dipole moment, all with regard to the normal (z-direction) of the plane spanned by the three sulfur atoms (x-y-plane).

Supplementary Note 5. SIMULATED FORCE CONSTANTS
The elastic deformation of a molecular junction is described to lowest order by Hooke's law, which states that the small displacements of the atoms from the equilibrium positions are proportional to the applied force. We simulated the compression or stretching of the molecular junction by slight changes (in steps of 0.02Å) of the position of the nitrogen atom in the z-direction. In the procedure, the sulfur atoms and the nitrogen atom were kept

Supplementary Note 6. SCANNING SIMULATIONS FOR BOND ENERGETICS
The z OFF -z ON -hysteresis map of spiro B is asymmetric with respect to a displacement of the tip along the molecular backbone, as shown in Supplementary Fig. 15. To inspect this, we calculated total energies for many positions of the STM gold tip in the y-z -plane, i.e., when it is placed above the backbone and in front of the molecular head as well as for different vertical distances from the molecule. In the simulations, we use a gold tip of 20 atoms and keep its two outermost gold layers fixed, defining the position of the tip. The system, consisting of tip and molecule, was subsequently relaxed by keeping, in addition to the outermost two Au layers, also the sulfur atoms fixed to mimic an infinitely stiff surface.
The total energy of the system was calculated after relaxation of the remaining free atoms of the tip and molecule. This procedure is justified by the fact that we observe essentially no hysteresis in the case that the three S atoms are fixed (see also Fig. 2c), and the energy is hence unique. Starting from a position close to the smallest total energy, the two outermost gold layers were moved in steps of 0.1Å horizontally either to the left (-y) or right (+y) or vertically down (-z ) or up (+z ). The total movement of the tip is 3Å along the y-axis and 2Å along the z -axis (see Supplementary Fig. 10).
In order to examine the effect of tip shape, we considered two configurations (see Sup- the experiment, we observe that the total energy maps reflect the asymmetry with respect to a displacement of the tip along the molecular backbone ( Supplementary Fig. 10a,b). The highest binding energies between tip and molecule occur in front of the nitrogen atom. The tip orientation does not affect this behavior. To quantify the binding, also the bond length between the Au tip atom and the nitrogen can be used. It shows a behavior similar to those of the total energy for both configurations ( Supplementary Fig. 10c,d).

Supplementary Note 7. DETAILS OF DFT+Σ CALCULATIONS
In this section, we present further details of our DFT+Σ calculations. We use them in order to better describe the quasiparticle energies and the level alignment in our metal- In the established approach, the screening of the metallic electrodes is described classically as the interaction of charge densities on the molecule, which we model through pointlike Mulliken charges, with two perfectly conducting semiinfinite surfaces. Following Ref.
[4], we use the charge distribution of the HOMO to calculate the image-charge correction, tively, as obtained from ∆SCF total energy calculations of uncharged and singly charged molecules [4].
Complications in the DFT+Σ procedure arise from the fact that we assume that the spiro molecules in the junctions are bonded to gold through thiolate bonds (see Fig. 3). Taking away the Au electrodes, the so-called "contacted molecule" in the junction thus forms a radical, since there are no hydrogen or acetyl groups attached to the three sulfur atoms of the tripod anymore. To determine HOMO and LUMO levels of a chemically inert molecule, we assume that the respective spiro molecule in the gas-phase features thiol SH endgroups at its legs instead (see Supplementary Fig. 8). We identify the HOMO and LUMO levels at the gas-phase molecule, and compute there ǫ H , ǫ L as well as IP, EA. Image charge corrections ∆ occ , ∆ virt are, however, computed at the contacted molecule, since the junction geometry enters. We have checked, as is visible in Supplementary Fig. 11   The submolecular structure of spiro B, consisting of two lobes that can be seen in constant height images (see Fig. 1d), agrees well with the simulated STM images (see Supplementary   Fig. 14a and Supplementary Fig. 12f). with the corresponding simulation further supports our proposed adsorption configuration (see Supplementary Fig. 14b). The ordered rows of spiro B (see Supplementary Fig. 14d) allow for a directional histogram that is presented in Supplementary Fig. 14c [110] [211] [101] [112] [011] [121] [110] [211] [101] [112] of spiro B that is shown in Fig. 4a and in Supplementary Figure 14a. The feedback loop was not used in this measurement, and the I-z curves start at a plane of constant height above the surface. The critical distance z ON , at which the contact is closed, is mapped in Supplementary Fig. 15a and reproduces the ellipsoidal areas of maximum apparent height of the molecule in the corresponding topographic image (see Fig. 1d and Supplementary   Fig. 14a). Supplementary Figure 15b shows the distances z OFF at which the contact reopens upon retraction of the tip. This map also displays a mirror symmetry with respect to the backbone of the molecule, but unlike Supplementary Fig. 15a it shows a monotonous variation along the axis of the backbone. This is also reflected in the map of the vertical hysteresis distance (see Supplementary Fig. 15c), that is, the difference between z OFF and z ON : Each molecule shows up as a characteristic c-shaped area of large hysteresis indicating a high binding energy between tip and molecule. This map of the hysteresis can be compared to simulations of the total energy which also reflect the asymmetry with respect to a displacement of the tip along the molecular backbone (see Supplementary Fig. 15d and Supplementary Figure 10 for details). At the rim of the ellipsoid, towards the center of the molecule (Supplementary Fig. 15e), the molecular junction largely stays closed upon retraction and the hysteresis is enhanced, while the bond breaks earlier and the hysteresis is reduced when the tip is placed laterally in front of the nitrile group ( Supplementary Figure 15f). This behavior is also observed for the electric-field induced contacting presented in Supplementary Figure 18 and is in full agreement with the proposed adsorption model and further emphasizes the need for a precise lateral control over the bond configuration in order to produce reliable molecular junctions. Note that molecules of the same orientation on the surface behave identically, which excludes a plastic reorganization of the junction and indicates a well-defined and reproducible situation which allows for a more detailed analysis of the energetics of bond formation and elastic molecular deformation. This is in agreement with previous studies which have shown that the nitrile terminal group is well-suited for highly reproducible contact formation to Au electrodes [6]. In contrast to the detailed study of contact formation between a metallic tip and a single CO molecule by Welker and Giessibl [7] and the experiments on forces during atomic manipulation [8], the gold-nitrile bond discussed here does not show a simple rotational symmetry of the energies involved in bond formation and molecular deformation but reproduces the reduced, twofold symmetry of the tilted molecular head group. Slight deviations from this characteristic shape seen on molecules of different orientation are probably caused by a non-symmetric arrangement of atoms at the tip apex.
Supplementary Note 11. POSITION DEPENDENCE OF THE ELECTRICALLY

INDUCED CONTACT FORMATION
As discussed in the main text, voltage-induced switching is only possible when the elastic energy needed to stretch the molecule is comparable to the energy of the tip-molecule bond.  Fig. 5a,b were recorded and analyzed at each pixel. The maps presented in Supplementary Fig. 17a-c were recorded at constant height above the substrate, and the voltage was swept at each pixel between −1 V and 2 V and back. The critical voltage that is needed to close the contact between tip and molecule (see Supplementary Fig. 17a) again reproduces the symmetric ellipsoidal shape of the molecular head group. In the center, even at a positive voltage (which effectively repels the molecular head group), the molecule closes the contact. We interpret this as a configuration at which tip and molecular head group are positioned at similar heights above the substrate. The higher sample voltage that is necessary to open the contact again shows a characteristic c-shape of lower symmetry than the voltage needed to close the contact (see Supplementary Fig. 17b). Similar to Supplementary Fig. 15c, a pronounced hysteresis is observed when the tip is positioned closer to the molecular center, while very low hysteresis is found when the tip is laterally positioned farther away from the center of the molecule (see Supplementary Fig. 17c and Supplementary Fig. 15e,f). The spatial distribution of the critical voltages for closing and opening the contact is clearly dictated by the orientation of the molecules so that an effect related to the shape of the tip apex [7] can be safely excluded. These observations of the spatial variation of the electrostatic contact formation are in full agreement with the experiments and simulations on the mechanical contact formation presented in Figs. 2,4. As can be seen in Supplementary Fig. 17e, the threshold voltage varies continuously with the position relative to the molecule. Finally, the conductance of the junction was derived from the I-V curves and is shown in Supplementary Fig. 17d. Again, each molecule shows up as a characteristic shape with a variation of the conductance along the axis of the molecular backbone. However, this variation is small compared to the difference between ON and OFF state conductances and, similar to the mechanically induced switching presented in Figs. 2,4, the corresponding conductance histograms shown in Supplementary Fig. 17f In analogy to the measurements on spiro A presented in the manuscript, the molecular junction formed with spiro B can be driven into a thermally bistable state by spatially positioning the tip and choosing the electric field appropriately. Again, we measured the energy difference between the two metastable states as a function of the applied bias voltage, that is, as a function of the applied electric field. Supplementary Figure 18a (see table S2). With this, the slope is given by Furthermore, we estimated the distance z tip between tip and sample surface from the tunneling current and the bias voltage. In the contact regime, the tip position can be identified with the height of the molecule. This value amounts to about 1100 pm in the case of spiro A and is slightly lower than the height of the molecule in the gas phase relaxation