Introduction

The concept of constructing devices exploiting the electronic properties of single molecules is promising. Organic molecules deposited on insulating substrates in the form of single or fractional (incomplete) monolayers are very attractive systems for making progress in this direction. As the electronic properties of a single organic molecule on a surface are determined by the way it is bound to the substrate, the control of molecule-to-substrate interaction and atomic scale positioning on the substrate is of central importance. Due to the small mass and small moment of inertia, thermal movements of the molecules and thermally activated rotation of the molecules on the surface can hinder the operation of a single molecule device. This raises the importance of understanding the binding, movement and rotation of organic molecules on substrates. We have recently studied a monolayer of C60 molecules deposited on surface of WO2/W(110) using ultra-high vacuum STM1,2,3. The molecules were found in conducting and nonconducting states, and our results indicated that such changes were associated with the molecule acquiring or losing a single electron, i.e. C60 on the surface could be charge-neutral, or it could gain/lose one electron. This effect can be understood in terms of Coulomb blockade blocking electron transport from the surface to the STM tip via the molecule. We have also established that when such a change in the conduction state of the molecules occurs, the molecule can switch between different orientations at the surface4. The change between static and spinning states of the molecules takes place at the temperature of 259 K. In a monolayer of C60, the change takes the form of a phase transition, similar to the one in the C60 crystal5,6,7,8, where the molecules of an entire monolayer switch from a static state to a spinning state2,3. In this paper, we investigate the nature of the electronic bonds of C60 molecules positioned on the atomically flat substrate when the molecules undergo the phase transition from the static to the spinning state. We demonstrate that at a certain temperature the molecules become loosely bound to the surface and at once initiate their rotational movement. This change in the binding energy between the molecule and the surface is accompanied by the molecule acquiring a charge and this can be controlled via STM tip bias. Sending electrons from the tip into the C60 (positive sample bias) activates molecular rotation while reversing the current direction (negative sample bias) suppresses the rotation. The results can be explained on the basis of Coulomb blockade charging of the molecule, leading to a change in the Coulomb interaction force between molecule and substrate.

Results and Discussion

A sub-monolayer film of C60 molecules deposited on the WO2/W(110) surface forms one molecule thick nano-islands, (Fig. 1a). Within the nanoislands the molecules are self-assembled into a close-packed hexagonal structure1 with 1 nm separation between C60 molecules, Fig. 1b.

Figure 1
figure 1

(a) 390 × 390 nm2 Low-temperature STM images acquired after the deposition of 0.5 ML of C60 molecules onto the WO2/W(110) surface. Vb = 1 V, It = 0.1 nA. (b) 13 × 10 nm2 Image of C60 at low temperature with the details of the sub-molecular structure. Vb = 0.97 V, It = 0.07 nA, T = 78 K. (c) 16 × 16 nm2 STM image of the same C60 film acquired at T = 315 K. All molecules in (c) appear as perfect spheres due their fast rotation. Vb = −1.4 V, It = 0.1 nA.

Full size image

At temperatures below 220 K, the molecules are completely static on the time scale of the STM experiment thus making it possible to reveal their sub-molecular electron orbital structure. The sub-molecular structure changes from one molecule to another as the molecules can be attached to the surface by 1, 2, 5 or 6 carbon atoms, known by rotational orientations a, h-p, h-h, p and h. Most molecules are found in the orientations corresponding to h-h attachment, see Fig. 1b. The specific attachment can be identified on the basis of comparison between density functional theory (DFT) calculations and the STM images2. Once the temperature is raised to 220 K, the molecule becomes unstable in any one binding position as the thermal energy of the molecule becomes comparable with the barrier separating the positions. Between temperatures of 220–260 K, the transitions between different molecular binding positions occur rarely enough so that the individual orientations can be time-resolved with STM and yet often enough so that changes could be observed during the time of the STM experiment. The transitions are best observed by positioning the STM tip above the molecule with either an open or closed tunnel current feedback loop. Figure 2 shows a molecule switching between five different states corresponding to five rational orientations of the molecule at the surface.

Figure 2
figure 2

A tunneling microscope tip is located above a fluctuating C60 molecule. Right panel: The time-evolution of the STM tip-surface distance for the switching C60 molecule measured at a sample bias, Vb = −1.1 V and tunneling current It = 0.087 nA. The acquisition time was 10 ms per point. Five different states are present indicated by dashed lines. Left panel: Histogram of C60 molecule residence in different states.

Full size image

The switching of the molecule between 5 rational orientations is detected as vertical height (z-position) change of the STM tip which can be seen in the right panel of Fig. 2. The current feedback remains on in this case. The transitions are manifested as changes in the z-position of the probe due to change of the C60 molecule’s orientation. The state where the molecule spends the longest time on average is the state with the lowest binding energy of the molecule. With reference to Fig. 2, this is concluded to be the h-h state. The majority of switching events observed in the experiment were due to the change of molecule orientation between the two lowest energy binding states. In Fig. 3a the STM probe is positioned directly above one molecule, the bias voltage is fixed at −1.1 V and the current feedback is disabled so that the tip height remains constant. The change in the tunnelling current due to the C60 molecule switching is observed in Fig. 3a. Such a mode of observation provides a faster response time of the STM. However, one could also operate the instrument with the feedback switched on and measure the z-position of the tip, see Fig. 3b. In any case, it is important that we now observe a transition between the ground state, with the molecule being fixed in one position, and the lowest energy excited state. The change in state of the molecule results in a change of current flowing through it. We shall discuss the details of this and the related mechanism later, for now we focus on the information that could be extracted from the mean time spent by the molecule in each state. Switching between two binding states of C60, which are adjacent in energy, is well described in terms of a two-valley model3. In this model, we consider two potential wells separated by an energy barrier U to estimate the molecule’s lifetime in two states, a depiction of this model is shown in Fig. 4a.

Figure 3
figure 3

The time evolution of the STM current (a) and tip-surface distance (b) for the switching C60 molecule between two lowest in energy orientations when the tunneling microscope tip is located above a fluctuating C60 molecule. Vb = −1.1 V.

Full size image
Figure 4
figure 4

(a) Model for two binding states of a C60 molecule. STM images of the C60 molecule before, during and after the switching are presented in inserts. (b) Dependence of potential barrier separating two adjacent in energy C60 molecule’s orientations on the applied bias voltage.

Full size image

The average lifetimes τ1 (τ2) of a molecule transitioning from state 1 (ground state) to state 2 and vice versa over a potential barrier separating the two states is determined by the number of attempts to overcome the potential barrier, i.e. Libron frequency fL:

$$\frac{1}{{\tau }_{1}}={f}_{L}\cdot {e}^{-\frac{U}{{k}_{b}T}}$$
(1)
$$\frac{1}{{\tau }_{2}}={f}_{L}\cdot {e}^{-\frac{U-\varepsilon }{{k}_{b}T}}$$
(2)

Here, U and U − ε are the heights of the potential barrier seen from the states 1 and 2 respectively. The difference in energies of the two states ε can be estimated from a statistical distribution of the residence time of the molecule in each state3. The difference in energies was estimated to be 18 ± 9 meV and does not depend on the bias which was used in the experiment. What is remarkable is that both potential barriers U and U − ε depend on the polarity of the bias. For positive bias, whereby electrons move from the tip to the molecule and then to the substrate, the energy of the potential barriers are some 80 ± 17 meV higher than in the opposing situation: negative bias, with electrons moving from the surface to the molecule and then to the tip. It is also notable that the energy of the molecule in the spinning state is not sensitive to the value of the bias voltage but rather to its polarity. The results shown in Fig. 4b represent statistics based on data collected from 35 molecules. Some scattering of the resulting values for the barrier in the plot could be due to C60 molecules being located at a number of non-equivalent sites on the WO2 surface.

At high temperatures, above the temperature of the rotational phase transition (TC = 259 K) the molecules rotate faster than the time scale of the STM experiment2. The sub-molecular structure can no longer be resolved and the molecules appear in STM as perfect spheres Fig. 1c. If the temperature is somewhat reduced, to around 256 K, we can establish a situation where fluctuations appear2,3 (see Supplementary video) and the C60 molecule typically switches between two states: the molecule being fixed on the surface, Fig. 5a and the same molecule rotating on the surface, Fig. 5b. A line profile indicated the height difference between the molecule and the substrate in this rotating state is presented in Fig. 5c.

Figure 5
figure 5

(a,b) Constant-current STM images (5.5 × 5.5 nm2) of the same area of the C60 monolayer on the WO2/W(110) surface, Vb = 1.0 V, It = 0.1 nA. The molecule at the center of these images switches between static and rotating states, changing its appearance. The molecule in (b) rotates faster than the time scale of the STM experiment. (c) A line profile (along the line marked in (b)) indicating the height difference between the rotating and static C60 molecules in the monolayer.

Full size image

In the first state, the molecule is static showing the sub-molecular features. This is the ground state as can be judged from the fact the molecule spends most of its time in this state. In the second state, the molecule spins faster than the time resolving capability of the STM. It, therefore, appears as a sphere in the STM image and the overall current through the molecule increases. The most surprising result was obtained when we measured the separation of the C60 molecules from the surface at different temperatures. The clearest way of doing this was to measure the height of the monolayer thick island of C60, seen in Fig. 6a by inspecting the step height from the substrate at the edge of the island. A sample line profile is shown in Fig. 6b. The results, presented in Fig. 6c, reveal an abrupt increase of the nano-island height at the temperature of the rotational phase transition (259 K) (indicated by the green line). Therefore, as the molecules start spinning at the surface they simultaneously move up and away from the surface and the bond between the molecule and the surface is altered. In the monolayer of C60 this takes the form of a well-defined phase transition at 259 K.

Figure 6
figure 6

(a) 160 × 160 nm2 STM image of nano-islands acquired after the deposition of 0.5 ML of C60 molecules onto the WO2/W(110) surface. Vb = 1.0 V, It = 0.1 nA. (b) A line profile (along the line marked in (a)) indicating the height of C60 nano-island. (c) Dependence of C60 nano-island height on the temperature. Abrupt increase of the height at temperature of rotation phase transition indicates break of C60 coordination bonds.

Full size image

The observed increase in the separation between the molecules and the surface of 1.2 Å is significant. It is in agreement with the difference in rotating and static molecule heights visible in the STM images and comparable with the difference in length between a coordinated and a van der Waals bond (0.9–1.2 Å)9,10,11. This value is much greater than the difference between the bond lengths for the five well-defined static positions of the molecule. The bond lengths obtained from the DFT calculations2 are presented in the table presented in (Table 1). The z-position for the ground state of the molecule (h-h orientation) is set as a reference point and the bonds of the other four orientations are given with respect to the h-h ground state position. The difference in the C60-WO2/W(110) gap for the five orientations is of the order of 0.5 Å or less, significantly smaller than the difference in the separation between the molecules and the surface across the rotational transition at 259 K.

Table 1 z-position (height above surface) of a C60 molecule’s center of mass for five rotational orientations with the ground state of the molecule (h-h orientation) set as a reference point.
Full size table

We have to conclude that during the rotation of the C60 molecule, the coordinated bond breaks and the molecule remains bonded to the surface via much weaker van der Waals forces, a schematic of this is presented in Fig. 7a. With this in mind, we can explain all the results presented here, including the abrupt dependence of the energy of the rotating state on the polarity of the bias voltage. The model of a molecule bonded to the surface by a weak van der Waals force is shown in Fig. 7b.

Figure 7
figure 7

(a) Schematics of C60 molecule bonded to WO2/W(110) surface by coordination (on the left) and van der Waals (on the right) forces. (b) Equivalent circuit of C60 molecule coupled in STM tunnelling junction. (c) Schematics of image charge of negatively charged C60.

Full size image

The model is characterised by two impedance values12: the impedance from the molecule to the surface (Rms) and the impedance from tip to the molecule (Rtm). The resistance Rtm has to be much greater than the value Rms, for the STM to operate in a stable regime13. In this case one could achieve the conditions for the Coulomb blockade if the electrons flow from the sample to the tip (negative bias) but not in the opposite regime: electrons flow from the tip to the sample (positive bias). Under the conditions of the Coulomb blockade, the C60 molecule can acquire and retain an extra electron. Once the bias is reversed, the molecule remains uncharged at any tunnelling current value. The charged molecule will create an image charge in the metal, seen in Fig. 7c, which results in a reduction of energy of the charged molecule by \(U=\frac{{e}^{2}}{2D}\), where D is the distance between the surface and the created charge. D can be estimated from DFT simulations (distance between molecule center and W atoms layer) and was taken to be about 8.7 Å. This results in an energy reduction of 414 meV. When the coordinated bond breaks and the molecule starts spinning, the change in D is ΔD = 1.2 Å. This results in a change in the energy barrier of ΔU = 50 mV, which is close to the experimental value of 80 meV. This deviation can be explained by 3 factors: (i) inhomogeneous charge distribution in a frame of the molecule4,14; (ii) there is some uncertainty in reference point of D; (iii) the electric field between the STM tip and the metal surface also contributes to the potential barrier. Therefore the results suggest that rotation state of the C60 molecule, which is realized at temperatures above that of the rotational phase transition, is different to any of the fixed binding states, as the energy of the spinning state is controlled by the bias of the tip and the direction of the current flowing through the molecule. The voltage controlling the switch between the static and the rotational states is around 1 V, the range convenient for typical electronic devices. The electronic impedance through the molecule is affected by the state of the molecule. This opens up the prospect of using such molecules on a surface as the core of an electronic current control device.

One could consider the analogy with a Kugel fountain, see Fig. 8. Once the granite sphere is lifted from the surface, it is easily rotated. When it comes down, it is positioned on the surface in a single well-defined position. This is broadly similar to the phase transition we observe for the C60 molecules on the surface. Once the coordinated bond breaks due to e.g. increased temperature, the molecule lifts away from the surface and it is then retained above the surface by a much weaker van der Waals force. In such a state the molecule is free to spin. The lifting of the molecule can be promoted or demoted by altering its charge. If the molecule is charged, it induces an image charge on the surface and this increases the energy binding the molecule to the surface and thus suppressed the transition from the bound state of the molecule to the spinning state. The charge of the molecule can activate the Coulomb blockade that is controlled by the direction of the current (into the surface or into the tip) and therefore controlled by the polarity of the bias voltage. Using the analogy of the Kugel fountain, this is similar to regulating the water pressure underneath the granite sphere: if the pressure is reduced, the sphere’s ability to go into the rotating state is suppressed. The observed difference in the potential barrier separating two adjacent in energy C60 molecule orientations, when the applied bias voltage changes polarity opens a way to control the rotation of the C60 molecules.

Figure 8
figure 8

Kugel fountain as a model of rotating C60 molecule. Image courtesy of M. Grassick.

Full size image

An STM image, Fig. 9a, acquired close to the rotational phase transition point at positive bias voltage (Vb = 1.2 V) is noisy and does not reveal the internal structure of molecules due to frequent switching of the molecules or their rotation. However, an STM image, presented in Fig. 9b, of the same area at Vb = −1.9 V demonstrates prominent internal structure of C60 molecules, which corresponds to molecular symmetry.

Figure 9
figure 9

Constant-current STM images (14 × 14 nm2) of the same area of the C60 monolayer on the WO2/W(110) surface, It = 0.1 nA (a) Vb = 1.2 V; (b) Vb = −1.9 V.

Full size image

Conclusion

We have studied the transition of C60 molecules deposited on a WO2/W(110) substrate from a fixed state, where the molecules are bound to the surface by a well-defined bond, to another state, where the molecules are loosely bound to the surface by van der Waals forces. The integrity of the monolayer of C60 molecules is not altered once the molecules transition to the loosely bound spinning state, but remarkably, the separation between the molecules and surface is increased by 1.2 Å across the transition. In the loosely bound state, the molecules start spinning around faster than the time resolving capability of the STM and therefore they appear in the STM as perfect spheres. In contrast, at low temperature, the molecules are bound to the surface in a static manner, and details of the sub-molecular orbital structure can be resolved with STM. The transition from the static state into the rotating state takes place at 259 K. The energy of the spinning state with respect to the lowest energy state having coordinated bonds can be obtained from statistics of the molecule’s switching. The binding energy of the molecule in the spinning state can be easily altered by changing the polarity of the bias voltage applied between the STM tip and the surface. The binding energy is reduced by 80 meV when the bias polarity of the sample changes from positive to negative with respect to the tip. The results are consistent with the Coulomb blockade model: when electrons travel from the surface to the C60 molecule and then to the tip, they produce an accumulation of charge on the molecule due to the Coulomb blockade. This increases the electrostatic interaction between the molecule’s charge and a corresponding image charge created on the metallic surface. The energy of this electrostatic interaction is consistent with the energy difference of the spinning state for positive and negative bias voltage polarity.

Methods

The sample preparation and experimental conditions closely match that of our previous work with C60 on WO2/W(110)1. The STM/STS experiments were performed using a commercial instrument from Createc, in an ultra-high-vacuum (UHV) system consisting of an analysis chamber (with a base pressure of 2 · 10−11 mbar) and a preparation chamber (5 · 10−11 mbar). An electrochemically-etched tip made out of tungsten single crystal with [100] crystallographic direction along the tip length, was used to record STM images in constant current mode. The voltage Vb corresponds to the sample bias with respect to the tip. No drift corrections have been applied to any of the STM images presented in this paper. A W(110) single crystal, prepared at the Institute of Solid State Physics RAS, was used as the substrate. An atomically-clean W(110) surface was prepared by in situ annealing at 1900 K in an oxygen atmosphere of 1 · 10−7 mbar, followed by a series of high temperature flashes at 2200 K. The sample was heated by electron beam bombardment and its temperature was measured using an optical pyrometer (Ircon UX20P, emissivity 0.35). The condition of the W(110) surface was verified by LEED and STM before oxidation. Once a clean surface was obtained, the sample was oxidised at 1600 K in an oxygen atmosphere of 1 · 10−6 mbar for 15 min. The quality of the resulting oxide structure was verified by Low Energy Electron Diffraction (LEED) and STM before the deposition of C60 molecules. C60 (Aldrich Chemicals) was evaporated in the preparation chamber isolated from the STM chamber at a rate of about 0.2 ML (monolayer) per min from a deposition cell operated at a temperature of approximately 700 K. Before evaporation, the C60 powder was degassed for about 8 hrs to remove water vapour. The total pressure during C60 deposition was in the 10−9 mbar range and the substrate was kept at room temperature. After deposition the sample was transferred into the STM and initially cooled down to 78 K before the temperature is raised in order to observe and conduct experiments relating to the rotational phase transition of the C60 molecules.