Tim
Kühne
ab,
Kwan Ho
Au-Yeung
ab,
Frank
Eisenhut
ab,
Oumaima
Aiboudi
cd,
Dmitry A.
Ryndyk
b,
Gianaurelio
Cuniberti
b,
Franziska
Lissel
*cd and
Francesca
Moresco
*a
aCenter for Advancing Electronics Dresden, TU Dresden, 01062 Dresden, Germany. E-mail: francesca.moresco@tu-dresden.de
bInstitute for Materials Science, TU Dresden, 01062 Dresden, Germany
cLeibniz Institute of Polymer Research, 01069 Dresden, Germany. E-mail: lissel@ipfdd.de
dFaculty of Chemistry and Food Chemistry, TU Dresden, 01062 Dresden, Germany
First published on 11th December 2020
Among the different mechanisms that can be used to drive a molecule on a surface by the tip of a scanning tunneling microscope at low temperature, we used voltage pulses to move azulene-based single molecules and nanostructures on Au(111). Upon evaporation, the molecules partially cleave and form metallo-organic dimers while single molecules are very scarce, as confirmed by simulations. By applying voltage pulses to the different structures under similar conditions, we observe that only one type of dimer can be controllably driven on the surface, which has the lowest dipole moment of all investigated structures. Experiments under different bias and tip height conditions reveal that the electric field is the main driving force of the directed motion. We discuss the different observed structures and their movement properties with respect to their dipole moment and charge distribution on the surface.
The influence of electric fields in STM manipulation has been well discussed for several decades.13,14 Recently, the molecular dipole moment has been reported to enable the controlled movement of molecules and nanocars using the external electrical field of the STM tip.9,15
This work aims at studying the lateral movement of di-tert-butyl-2-isocyanoazulene-1,3-dicarboxylate (BCA) single molecules and nanostructures on the Au(111) surface induced by the inelastic electron tunneling and electric field effect. The molecules were specifically designed to have a high molecular dipole moment while decoupling the azulene core from the Au substrate by two tert-butyl groups. After deposition on the surface, we observed single adsorbed molecules and self-assembled nanostructures with different dipole moments. We investigated under the same experimental conditions their non-contact manipulation, studying the role of the dipole moment and charge distribution on their lateral movement.
The target di-tert-butyl-2-isocyanoazulene-1,3-dicarboxyl (BCA) (5) was obtained in four steps starting from commercially available tropolone in an overall yield of 58.2% and is stable under ambient conditions. Tosylated tropolone (1) was obtained from tropolone in a near quantitative reaction following a literature protocol.19 The subsequent formation of di-alkyl 2-hydroxyazulene-1,3-dicarboxylate is usually achieved via a procedure by Nozoe et al.20,21 using an active methylene reagent in the presence of a base in ethanol. Yet, following this protocol using tert-butyl cyanoacetate led to a base-catalyzed transesterification, yielding diethyl-2-aminoazulene-1,3-dicarboxylate instead of the targeted di-tert-butyl derivative (2). The conversion was finally achieved using THF at 50 °C, giving the di-tert-butyl-ester (2) in a good yield. Subsequent formylation22 of (2) with an excess of acetic-formic anhydride afforded (3) in 96% yield. Finally, the red colored formamide (3) was converted into the corresponding isocyanide (4) by dehydration22 with POCl3 in the presence of Hünig's base. The target compound (5) was obtained as a magenta solid in 84% yield and can be easily distinguished from the formamide on the basis of the characteristic isocyano signals in the 13C NMR (δ = 176.6 ppm in CDCl3, Fig. S14†) and FTIR (νNC = 2128 cm−1, Fig. S17†) spectra.
Scheme 1 Synthetic route for the preparation of di-tert-butyl-2-isocyanoazulene-1,3-dicarboxylate (4). |
When evaporating the same molecules maintaining the surface at room temperature, we observe small molecular assemblies in the STM images (Fig. 1b), while single molecules are very rare. Two distinct nanostructures are repeatedly present on the surface that both exist in two enantiomers and exhibit point symmetry: a nearly square dimer, visible at the bottom right corner of Fig. 1b (dimer 1, 2%, red contour), and a smaller dimer (dimer 2, 98%, bright blue contour). Both dimers were studied in more detail and their structure was determined with the help of density functional theory (DFT) and STM image calculations (see Methods for further details).
In Fig. 2, the STM image of dimer 1 (Fig. 2a) is compared with the adsorption geometry optimized by DFT (Fig. 2b) and with the calculated STM image (Fig. 2c). Considering the good agreement between the calculated and experimental images, we can conclude that dimer 1 is composed of two intact molecules forming a metal–organic complex with two Au adatoms. This conclusion is supported by the experimental height profiles of Fig. 2d, comparing a single molecule with dimer 1. The best agreement between theory and experiment occurs for a metal–organic complex where two gold adatoms lie in the centre, thereby stabilizing the structure and lowering the energy by about 0.5 eV. The formation of such metal–organic complexes with diffusing metal adatoms on metal surfaces is a well-known process on several metal surfaces including gold.23 This process can be enhanced by the cyano-functionality, which, as reported recently, can also be used for 2D metal–organic network assemblies.24,25
Fig. 3 shows a STM image of dimer 2 (Fig. 3a) and the calculated structure (Fig. 3b) and the corresponding simulated STM image (Fig. 3c). By comparing theory and experiment, we assign this type of dimer to a metal–organic structure formed by two gold adatoms and two molecules, where on each one of the tert-butyl carboxylate groups was cleaved during evaporation. Similar to the case of dimer 1, the gold adatoms stabilize the structure. This result is confirmed by taking a single molecule evaporated on a cold surface and cleaving one of its tert-butyl groups by voltage pulses of 2.5 V (Fig. S7†). As one can see in Fig. 3d, the height profiles of dimer 2 and one of the cleaved single molecules are in good agreement.
The dipole moments of adsorbed molecules and nanostructures were determined by DFT, obtaining 9.0 D for the single molecule, 6.5 D for the symmetric dimer 1, and 2.3 D for the asymmetric dimer 2 (see Table S1 in the ESI†). The dipole moments are determined by the charge distributions over the atoms shown in Fig. 5(b) and (c). At a first glance, one might think that the dipole moment of dipole 2 is larger because of the larger charge separation. However, actually, this dipole moment is the smallest one because the dipole moments of the monomers are counter.
Fig. 5 (a) Threshold voltage for movement versus tip height above the surface during voltage pulses applied to dimer 2. The zero height corresponds to the experimentally determined tip–Au(111) point contact position. Error bars are due to threshold differences between the molecules, probably due to changes in the tip or in the adsorption position. The slope of the linear fit function (black line) is 0.36 ± 0.01 V Å−1. (b and c) Calculated charge distribution for dimers 1 and 2, respectively. (Refer to Fig. S8† for the side views.) |
To describe the observed behaviour, we can argue that the electric field deforms the potential energy surface for diffusion, thus leading to an effective lowering of the energy barrier. Inelastic tunneling electrons also contribute to the process providing further energy to overcome such a barrier.
In contrast to other studies, where molecular dipole moments are considered to be determinant for voltage pulse induced manipulation,15 neither the BCA single molecules nor dimer 1 could be moved by voltage pulses (and electric field) despite their high dipole moment of 9.0 D and 6.5 D, respectively. To understand this result, we went a step ahead calculating the charge distributions for dimers 1 and 2 on the Au(111) surface (Fig. 5b and c), which can provide more specific information than the dipole moment. For the intact molecules forming dimer 1 (Fig. 5b), the positively charged Au atoms in the centre of the nanostructure weaken the charge separation by compensating for the effects of isocyano and carboxyl groups. Oppositely, the isocyano and oxygen in the centre of dimer 2 (Fig. 5c) cause a comparatively larger charge separation. Here, the gold atoms are located at the sides of the structure, leading to a negatively charged centre encapsulated by positively charged parts of the complex. This higher charge separation might explain the stronger reaction to electric fields of dimer 2 with respect to dimer 1 and single molecules.
Considering the polarity dependent motion of dimer 2, we observed that, for a positive voltage pulse at the tip (i.e. a negative bias voltage), the negatively charged centre of dimer 2 is attracted by the tip, while it is repulsed by the opposite tip polarity. Therefore, when inducing the motion of dimer 2, a stronger charge separation, but not necessary a higher dipole moment, may explain the observed bias selective directional behaviour.
A Au(111) single crystal was used as the substrate. It was cleaned by subsequent cycles of Ar+ sputtering and annealing to 723 K. The BCA molecules were evaporated on the sample surface maintained either at room temperature or at 77 K using liquid nitrogen cooling to prevent molecular diffusion on the surface. For the 30 s sublimation, a Knudsen cell was heated to 380 K. Both sample preparation and experiments were performed under ultra-high vacuum (UHV) conditions at low 10−10 mbar pressure. STM measurements were performed at 5 K sample temperature. All images were obtained in the constant current mode at a high tunneling resistance in the order of 10 GΩ to prevent accidental excitation. The bias voltage was applied to the sample with respect to the tip.
Voltage pulse manipulation was performed by positioning the tip near to the molecule (see the marks in Fig. 4) at a specific height with respect to the bare Au(111) surface and applying a selected bias voltage for 1 s. The height was calibrated by recording I(z) curves. Voltage pulses were applied at a constant tip height. I(t) curves were recorded during the pulse, detecting the movement of the molecule normally after about 0.1 s (Fig. S5†). STM images were recorded before and after the application of the pulse, determining the displacement of the molecule.
The calculations of molecular geometry, charge distributions, and STM topography images were performed using TraNaS OpenSuite (tranas.org/opensuite), partially based on the DFTB+26,27 software package and the CP2K software package (cp2k.org) with a Quickstep module.28 For geometry optimization, we used the DFT method as implemented in the CP2K package. We applied the Perdew–Burke–Ernzerhof exchange–correlation functionals,29 the Goedecker–Teter–Hutter pseudo-potentials30 and the valence double-ζ basis sets. We also used the density functional based tight-binding method with auorg-1-1 parametrization31,32 as implemented in the DFTB+ package. For geometry calculations, we used the DFT-D3 method of Grimme33 for the van der Waals correction. As the charge density distributions and STM images were calculated by the DFTB method, we compared the DTFB and DFT methods for geometry optimization and electronic properties of molecules in the gas phase, and found perfect agreement.
To simulate the STM images and charge distributions, we considered a realistic atomistic system including the STM tip and the substrate, both connected to semi-infinite electrodes. The electrodes have been modelled by a periodic slab of triad layers of Au(111). The tip consists of three layers of Au(111) with sixteen atoms in a pyramidal shape with one atom on the apex, which is attached to the underside of the top electrode. The simulation of STM images in the constant-current mode was performed based on the current calculations by the Green function technique.34 It is important to note that in our approach the atomic structure of a surface and a tip and possible interference effects between substrate/tip and substrate/molecule/tip currents are taken into account automatically and the algorithm of the tip motion reproduces the experimental one. The data are analyzed and the images are obtained using the PyMOL Molecular Graphics System, Version 2.4 open-source build, Schrödinger, LLC.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d0nr06809h |
This journal is © The Royal Society of Chemistry 2020 |