Noncontact Atomic Force Microscopy

NCAFM experiments were performed by A. Amrous, F. Bocquet, L. Nony, F. Para, and C.

Loppacher at Aix-Marseille University. The images shown in this chapter were acquired at room temperature using a commercial AFM (VT-AFM, Omicron Nanotechnology GmbH 65232 Taunusstein, Germany) equipped with home-built electronics and an RHK controller (SPM 1000, PLL Pro II, RHK Technology Inc. Troy, MI 48083 USA). The cantilevers used

in this chapter were Nanosensors (PPP-NCL, resonance frequency 150 kHz, spring constant 30 N/m, quality factor Q approximately 40000) operated at typical oscillation amplitudes A₀ of 10 nm.

The single crystal KCl substrate (MaTecK GmbH, 52428 Jlich, Germany) was cleaved ex situ, quickly introduced into UHV, and annealed at approximately 200^◦C for 1 hour in order to produce atomically clean substrates with large terraces. CDB molecules (and variations) were deposited using home-built crucibles at evaporation rates of approximately 1 ML (monolayer) per minute onto substrates kept at room temperature.

CDB molecules deposited onto KCl(001) were imaged using NCAFM at room temper-ature. Therefore, it was impossible to resolve single molecules on the surface. Instead, the images collected show a porous mesh-like structure as shown in Figure 6.2A. A closer look at the mesh structure reveals a periodic square network tilted approximately 8^◦ with respect to the <100> direction of the surface, as shown in Figure 6.2B. While the physical meaning of the bright and dark spots in these images is unclear, the experimental data provided me with the periodicity and orientation of the monolayer structure with respect to the KCl(001) surface.

Figure 6.2: CDB molecules on KCl(001), A) NCAFM large scale topography image showing two different molecular domains separated by a monatomic step of the substrate. B) A molecular scale image and an inset indicating the orientation of the substrate with ionic resolution acquired on different areas of the substrate (inset frame edge=4 nm).

My initial task was to propose atomistic structures for the CDB network and select a model of the NCAFM tip in order to produce VAFM images that could be compared directly to these results. In order to do so, I had to accurately represent the molecule-surface interactions using a method that could represent the system periodically while constraining the CDB layer as little as possible.

6.3 Theoretical Methods

In this study I employed a multi-scale approach to study the layer structure of CDB molecules on KCl(001). DFT calculations were used to characterize the adsorption of single CDB molecules on the KCl(001) surface and compare the various competing inter-actions present in this system. I then used the information gained to predict monolayer structures and calculate their energies using the previously discussed periodic QM/MM scheme. Finally, I applied the point dipole model discussed in Chapter 5 to simulate a VAFM image that can be directly compared to experimental data.

6.3.1 Density Functional Theory

I performed DFT simulations to characterize CDB adsorption on KCl(001) and compare the relative strength of the competing molecule-molecule and molecule-surface interactions in this system. These calculations were made using the mixed Gaussian and plane waves (GPW) [13] method as implemented in the CP2K code. In this work I was able to properly represent the electronic structure of the surface with the PBE GGA density functional and the MOLOPT basis set [19], avoiding the need to use a computationally expensive hybrid functional. The HOMO-LUMO gap of bulk KCl(001) was calculated to be 5.6 eV using this method which can be compared to an experimental value of 7.6 eV [35].

The surface was represented using four atomic layers with one fixed layer at the bottom and three layers of relaxed atoms in order to reproduce the band gap, rumpling, and lattice constants. I employed semi-empirical long-range dispersion corrections [176] in order to

represent vdW interactions in the system, as described in Chapter 2. Finally, I chose the MOLOPT basis set [19] to minimize basis set superposition error (BSSE). Using these methods, the HOMO-LUMO gap of the KCl(001) surface was calculated to be 5.4 eV with a lattice constant of 6.3 ˚A. Surface K cations were displaced 0.03 ˚A into the surface plane, while Cl anions protrude 0.03 ˚A above the surface plane. These results are in line with experimental low energy electron diffraction (LEED) studies that observed a lattice constant of about 6.3 ˚A and surface rumpling on the order of 0.03 ˚A [36].

My DFT calculations included geometry minimizations of single CDB molecules and several variations of the molecule adsorbed onto KCl(001). The lowest energy configuration of each molecule was computed by placing the molecule onto the surface in a number of starting positions and optimizing the geometry. These starting positions were obtained by positioning the molecule so that the molecular board was parallel to the surface plane and rotating the molecule by 2^◦ increments. This was repeated several times with the center of mass of the molecule above a cation, an anion, and a bridging site on the surface. I also studied the competing interactions within the system. In order to accomplish this, the molecule was divided into several main components by breaking bonds in order to examine the various contributions to interaction energy between molecules or adsorption energy on the KCl(001) surface. Each of these broken bonds was terminated with a H atom. The fragments studied include A) a hexane molecule representing the hydrocarbon chains, B) a benzonitrile molecule representing the functional groups of the larger CDB molecule, and C) a benzene molecule representing the core of the CDB molecule. Examples files for studying the CDB molecule on KCl using CP2K can be found on the supplementary disk attached to this thesis.