Low coverage deposits
An image of the molecules at low coverage, deposited at room temperature on KBr(001), is shown in . The white dots that appear on the step edge have a size that is compatible with single molecules, but the resolution is not high enough for a convincing identification. The dots on the terraces are more extended. Their diameter ranges from 3 to 5 nm whereas their height reaches 0.9 nm. We interpret them to be small molecular aggregates comprising a few to a few tens of molecules.
Constant-frequency-shift image of the KBr surface after the deposition of a small amount of molecules at room temperature. Imaging conditions: Δf = −5 Hz, oscillation amplitude A = 2 nm.
A KBr terrace with a higher coverage is shown in , with its simultaneously measured Kelvin voltage map. A negative shift of approximately 0.4 V appears on the largest aggregates relative to the mean KBr signal. According to the standard interpretation [28
], the sign of this shift is indicative of the presence of permanent dipoles that have a component pointing outward from the surface, or of positive charges under the tip.
Upper image: topography and lower image: Kelvin map of a KBr terrace with a higher coverage. Imaging conditions: Δf = −20 Hz, A = 2 nm. The profiles correspond to the white lines drawn on the images.
Higher coverage deposits
For higher coverage, the molecules were deposited at room temperature and the surface was studied at room temperature and after annealing at 80 °C and 150 °C. The complete study of the evolution of this system with temperature is not the purpose of this report. Here, for the sake of simplicity, we only discuss the results after annealing at 150 °C. Note that the types of structure we observed after the 150 °C annealing were already present at room temperature, differing essentially by the size of their domains. The images of a higher-coverage deposit, annealed at 150 °C during 30 min, presented in , show that several structures coexist on the surface.
Figure 4 Images of a sample annealed at 150 °C after the deposition of the molecules at room temperature. (a) and (b): topographic images, (c) and (d): enlargement of the areas marked by a white rectangle on (a) and (b); (e) and (f): Kelvin maps obtained (more ...)
We first focus on the upper part of the images. It can be seen by comparing and that the line bordering the large triangular area on its right (arrows in and ) has been displaced toward the right during the 13 h time lapse separating them. The enlargements of and show that the surface liberated by this process presents dots that are quite similar to the molecular aggregates of . For this reason, we identify this region as the KBr substrate. The phenomenon observed in can then be attributed to a dewetting process of a molecular layer [4
] corresponding to the domain labeled MLh in . Its height of 0.4 nm (profile in ) is compatible with the height of a molecule lying flat on the substrate. Observation of such an evolution at room temperature on a system that has been annealed at 150 °C shows that it is kept far from equilibrium by the very slow kinetics of reorganization. The Kelvin maps of and show a very clear contrast between KBr and MLh (profile in ). The Kelvin bias on MLh is shifted toward negative values relative to KBr, that is, in the same direction as for the aggregates discussed previously.
(a) and (b) are the profiles that correspond to the blue and green lines drawn in ; (c) is the profile corresponding to the black line on , and (d) is the histogram of .
We now examine the third type of domain present in these images (labeled MLv), which displays dark lines oriented along a polar <110> direction (see also ). The profiles of and show that their height relative to the KBr surface amounts to approximately 1.6 nm. This value does not vary from one layer to the other (note that the domain located in the lower part of and is crossed by a KBr monoatomic step, outlined in ) and is comparable to the diameter of a molecule (see the scale in ). This observation indicates that these MLv domains comprise a layer of molecules standing approximately upright on the KBr surface. The mean distance between the dark lines is around 11 nm and the associated corrugation is between 0.1 and 0.2 nm. The Kelvin maps of and show that the Kelvin signal is also shifted toward negative values relative to KBr for the MLv domains, with a comparable mean Kelvin bias (histogram in ). Nevertheless, the Kelvin map on the two types of structures has a different aspect, being more heterogeneous on the MLv domains. This heterogeneity is not clearly correlated to the topographic images.
High resolution (a) topographic and (b) Kelvin image of an MLv domain. A = 2 nm, Δf = −20 Hz. The arrows in (a) point to the ~4 nm modulation.
Another example of images of a high-coverage deposit, annealed to 150 °C, is shown in . Some well-resolved defects appear in the MLh domain in the upper-left area of the topographic image of (see the enlargement in ). These defects are also visible in the Kelvin map of (enlarged in ). They appear as clear dots, corresponding to a positive shift of the Kelvin voltage on the order of 0.8 V relative to the mean Kelvin voltage of the surrounding MLh domain. As expected the spatial resolution in the Kelvin map is lower than in the topography map due to the longer range of electrical forces relative to van der Waals forces.
(a) Topographic and (b) Kelvin map of a high molecular coverage annealed to 150 °C; (c) and (d) are enlargements of the upper-left parts of (a) and (b) (blue rectangle). Δf = −20 Hz, A = 2 nm.
The different domains that appear in have been labeled and the monoatomic KBr steps outlined in green. These attributions are based on the measurement of the height of the different structures and their Kelvin signature, as discussed previously. The steps have a remarkable shape, quite different from what is observed on the clean KBr surface, before adsorption, where they are mostly straight and aligned along the nonpolar KBr(001) directions. It is clearly seen that the step morphology is strongly coupled to the structure of the MLv domains. The steps tend to align along the same polar direction as the dark lines of the structure. These steps are highly unstable on the clean surface due to their high electrostatic energy. They can be stabilized only by charged species, in the present case by adsorption of the negatively charged N atoms of the CN groups. This observation also points to a massive KBr surface mass transfer during the annealing of the substrate due to molecular adsorption. The mechanisms at work during this transformation could be of the same nature as those discussed recently in the study of the restructuring of KBr(001) steps by truxene molecules [11
]. Finally, we note that when the molecular structure crosses portions of steps that are not aligned in these directions, these dark lines are not visibly affected, indicating that this structure has a strong intrinsic cohesion.
High-resolution images of MLh and MLv domains
Two high-resolution images obtained on the same MLh domain are shown in . The molecular network can be described by a unit cell characterized by u (2.9 nm, −13° from ) and v (3.7 nm, +61° from ) (). Note that due to the different imaging conditions in and , the molecular layer appears as a network of black holes in (a) and white bumps in (b). Comparing the size of this unit cell with the size of the molecule () suggests that the basis of the network comprises two molecules.
High resolution topographic images of an MLh domain. A = 2 nm. (a) Δf = −35 Hz, (b) Δf = −50 Hz. The unit cell is indicated in (b).
The images on a MLv domain displayed in show that the dark lines that appear in the large-scale images of and are separated by thinner lines, delimiting rows with a width of ~2.3 nm, slightly larger than a single molecule. A modulation with a period of ~4 nm appears along the rows. Because of the above-mentioned observation that the dark lines can cross a KBr atomic step without be perturbed, we tentatively interpret these observations as indicating that one row corresponds to a stack of molecules in relatively strong interaction. Note that, as remarked before, the Kelvin map () is very heterogeneous, with values of the Kelvin voltage varying between −0.5 and +1.2 V. This dispersion is indicative of a certain degree of disorder as is also observable in the topography image of .
To gain insight into the adsorption and dynamic properties of HCPTP on KBr(001), we performed numerical simulations, as described in the Methods section. The calculated lowest-energy adsorbed conformation of HCPTP on KBr(001) is displayed in .
Lowest-energy adsorbed conformation of HCPTP adsorbed on KBr(001). (a) Top and (b) side view. K+ ions are violet, N atoms are blue. The arrow in (b) points to the CN group that is not bound to a K+ ion.
The molecule is bound to the surface by the electrostatic interaction between its CN groups and K+
ions. The flexibility of the propyl chains allows the molecule to reach five K+
ions. One of the chains cannot bind and its CN group stays at a larger distance from the surface (). The N atoms of the CN groups that bind the molecule are at a mean distance of 0.28 nm while the central aromatic core lies flat at a distance of 0.4 nm from the surface plane. The calculated adsorption energy of 1.8 eV is quite large. It includes not only the contribution of the five CN groups but also the interaction energy of the negatively charged oxygen atoms and the aromatic core with the surface, which can be roughly evaluated by calculating the adsorption energy of hexamethoxytriphenylene on KBr(001) under the same conditions. We obtain 0.8 eV, meaning that each CN group contributes approximately (1.8 − 0.8)/5 = 0.2 eV, in good agreement with the value obtained for the CN groups of the truxene derivative mentioned previously [10
Molecular dynamics studies of the diffusion of a HCPTP molecule were performed with the same force field in the NVT ensemble with a Nose–Hoover thermostat. Simulations at 300 K show that the molecules diffuse by successive hopping of CN groups from one K+
to another in a way that is similar to the "walking" of the truxene-derived molecule described recently [10
]. To get an order of magnitude for the diffusion coefficient, we observe that the molecule travels approximately 1 nm in a time T
= 2.5 ns. Thus, D
) ≈ 10−10