Steering Self‐Assembly of Three‐Dimensional Iptycenes on Au(111) by Tuning Molecule‐Surface Interactions

Abstract Self‐assembly of three‐dimensional molecules is scarcely studied on surfaces. Their modes of adsorption can exhibit far greater variability compared to (nearly) planar molecules that adsorb mostly flat on surfaces. This additional degree of freedom can have decisive consequences for the expression of intermolecular binding motifs, hence the formation of supramolecular structures. The determining molecule‐surface interactions can be widely tuned, thereby providing a new powerful lever for crystal engineering in two dimensions. Here, we study the self‐assembly of triptycene derivatives with anthracene blades on Au(111) by Scanning Tunneling Microscopy, Near Edge X‐ray Absorption Fine Structure and Density Functional Theory. The impact of molecule‐surface interactions was experimentally tested by comparing pristine with iodine‐passivated Au(111) surfaces. Thereby, we observed a fundamental change of the adsorption mode that triggered self‐assembly of an entirely different structure.

The surfaces were checked by STM imaging, where emergence of an unperturbed herringbone reconstruction indicated their cleanliness.
Iodine-passivated Au(111) (I-Au(111)) samples were prepared by exposing freshly prepared Au(111) substrates to I2 vapour (~910 -6 mbar) for ~20 min in a separate preparation chamber to avoid cross-contamination of the main chamber. Thereby, holding the Au(111) surface at an elevated temperature of ~200 °C and cooling down in the I2 atmosphere aided in obtaining a completely closed iodine monolayer devoid of vacancies. Alternatively, samples were annealed to ~150°C after the iodination to form the monolayer.
We used a home-built low-temperature STM operated by a BP4 control system (Nanonis) for sample characterization. STM imaging was carried out at ~90 K using electrochemically etched tungsten tips that were in-situ conditioned by Ar + -ion sputtering. Stated tunneling voltages refer to the sample. STM images were processed by levelling and mean filtering.

NEXAFS experiments
NEXAFS experiments were carried out at BESSY II synchrotron (Helmholtz-Zentrum Berlin) at the HE-SGM beamline. Fantrip on pristine Au(111) samples were prepared as outlined above by in-situ deposition. Yet, iodination of Au(111) samples was not possible at the beamline.
Hence, we prepared I-Au(111) samples in advance in our home laboratory. These samples were characterized by STM, stored under vacuum and swiftly transferred through atmosphere to UHV at the synchrotron. For these experiments we used Au(111) films on mica (Georg Albert PVD-Beschichtungen) as substrates.
Carbon K-edge NEXAFS spectra were acquired for incidence angles of 30°, 45°, 55°, 70° and 90° (=normal incidence) with respect to the surface plane. A home-built double channel plate detector was used in partial electron yield (PEY) mode, i.e. with a counter voltage of -150 V applied. Photon energies were calibrated by means of an internal carbon reference. Spectra of clean Au(111) were used for both background and photon flux correction.
Intensities of the C 1s  * resonances were determined as the peak maximum values.
Theoretical curves were calculated for the (111) surfaces by using the standard equation for three-fold symmetric surfaces, [1] and by considering the beamline-specific degree of linear polarization of P=0.92.

DFT calculations
Periodic Density Functional Theory (DFT) calculations were performed with the VASP code, [2] using the projector-augmented wave method to describe ion-core interaction. [3] Exchangecorrelation were described by the van der Waals density functional (vdW-DF), [4] in the recent form by Hamada, [5] denoted by rev-vdW-DF2, which has been shown to accurately describe molecular adsorption in a variety of systems. [5][6] The Au(111) surface was represented by a slab of four layers. We used a (5√3 × 5√3)R30° surface unit cell (with respect to the primitive Au(111) unit cell) for all calculations of the adsorption of isolated fantrip molecules on both pristine and iodine-passivated Au(111). The iodine monolayer was modelled as (√3 × √3) 30° superstructure with iodine atoms adsorbed in three-fold hcp-sites in accord with literature. [7] For all adsorption configurations, we first performed calculations with a Γ-point only For calculations of free-standing fantrip and antrip monolayers, a 450 eV kinetic energy cutoff was used to ensure convergence when comparing calculations that used different unit cell sizes. Structural optimizations were performed on all atomsexcept the bottom two Au layers of the Au slab, which were kept frozenuntil the residual forces were smaller than 0.01 eV/Å. STM image simulations were performed within the framework of the Tersoff-Hamann approximation, [8] using the implementation by Lorente and Persson. [9] Fig. S1 NEXAFS of fantrip on I-Au(111). a) Carbon K-edge spectra acquired for incidence angles between 30° and 90° (normal incidence); b) intensity plots derived from a); data points are shown as filled symbols; Filled triangles correspond to intensities from the first resonance at 285.0 eV, whereas filled circles correspond to the second resonance at 286.1 eV. The solid lines represent theoretical intensity plots computed for different dihedral angles  between Au(111) and anthracene blades, where =90° corresponds to a fully upright, i.e. edge-on orientation. Accordingly, edge-on adsorption of the anthracene blades is corroborated.

Fantrip on iodine-passivated Au(111)
Step-edges  Fig. S4 a) / b) STM images acquired after fantrip deposition onto cooled I-Au(111) held at a temperature of ~80 K. Thereby, fantrip diffusion, and consequently self-assembly were kinetically hindered, resulting in smaller dispersed aggregates of variable size rather than extended domains. We routinely observed dimers (red circle in a), often attached to stepedges) and round isolated features (blue circle in b)). Their sizes are compatible with a single fantrip molecule with its anthracene blades adsorbed edge-on. Although we anticipate smearing out of internal contrast by thermal motion, the assignment is not unambiguous.  Fig. S5 a) c) STM images acquired after antrip deposition onto pristine Au(111). Analogous to fantrip, each protrusion corresponds to a single antrip molecule adsorbed with two anthracene blades parallel to Au(111) in accord with the scaled overlay in c). In contrast to fantrip, the monolayer is comprised of antrip dimers that are periodically arranged and separated by small gaps. The difference to fantrip is readily explained by the absence of intermolecular hydrogen bonds, as antrip lacks the fluorine substituents, hence an hydrogen bond acceptor. Therefore, molecule-surface interactions become more important and presumably imprint this irregularity of intermolecular spacings. Moreover, it is also possible to fit antrip molecules with edge-on adsorbed anthracenes into the gaps as shown in b). This offers an alternative explanation for the gaps, where the upright antrip molecules additionally stabilize the structure by - interactions in a T-shape configuration. (tunneling parameters and scale bars: a) -1.00 V; 5 pA, 20 nm; b) -1.00 V; 5 pA, 4nm; c) +1.00 V; 5 pA, 4nm) Fig. S6 a) overview and b) close-up STM images acquired after antrip deposition onto I-Au(111). Antrip self-assembles into a hexagonal porous monolayer with a lattice parameter of a = b = 1.95 ± 0.10 nm. The antrip and fantrip monolayers on I-Au(111) are isostructural (Fig.  4 of main manuscript). This implies antrip adsorption with all anthracene blades edge-on and face-to-face stacking as indicated by the overlay in the lower right corner of b). Remarkably, the antrip structure on I-Au(111) also adopts the same commensurate 44 superstructure (cf.   vacuum and b) fantrip molecules in the hexagonal free-standing monolayer (blue (red) corresponds to negative (positive) electrostatic potential). As expected, the partial negative charges acquired by the highly electronegative fluorine-substituents give rise to a negative electrostatic potential. The face-to-face stacking in the monolayer packing features an attractive electrostatic interaction with the slightly positive hydrogen-substituents. Yet, the monolayer packing also acts back on fantrip's charge distribution. This becomes evident in c) the charge difference plot for monolayer formation with reference to isolated fantrip molecules in vacuum (blue (red) corresponds to electron accumulation (depletion)). The fluorine-substituents acquire more negative charge in the monolayer packing, while electrons are shifted away from the almost perpendicular anthracene blades at the side to which they point. This results in a strengthening of - interactions between anthracene blades in the Tshaped configuration (highlighted by black circles in b) and c)).       Fig. S11c), where fantrip adsorbs with its anthracene blades edge-on aligned in the troughs of the iodine monolayer. a) Charge difference plot (blue (red) corresponds to electron accumulation (depletion)) for adsorption with reference to isolated molecules in vacuum and an unperturbed I-Au(111) surface, respectively (iodine atoms are depicted as purple spheres, the underlying Au(111) has been omitted for clarity). The plot exemplifies that negative charge is shifted away from the iodine atoms underneath fantrip's highly electronegative fluorine-substituents toward the center of the molecule. b) Bader charge analysis of the iodine atoms in the monolayer. For the pristine I-Au(111) surface chemisorbed iodine atoms acquire a minor negative charge of -0.078 e. Evidently, the changes in the proximity of the fluorine-substituents are extremely small. Yet, even the most affected iodine atoms still remain negatively charged, ruling out a favourable electrostatic interaction between fluorine and iodine.  Fig. S11 (dark blue line). For this comparison, the vacuum levels for both systems have been aligned. The differences remain astonishingly insignificant, clearly indicating that fantrip's electronic structure is not significantly altered upon adsorption. This is an unambiguous signature for physisorption on the weakly interacting I-Au(111) surface. S13

STM image simulations
Fig. S16 STM image simulations of a hexagonal fantrip monolayer adsorbed on I-Au(111). The monolayer structure is based on the experimentally observed 44 superstructure and the DFToptimized adsorption site of single fantrip molecules (cf. Fig. S11 c)). Electronic states were evaluated in the energy range between Fermi level (EF) and a) -2.0 eV to h) +2.0 eV as indicated. The overlaid structures in the lower left corners are meant to provide orientation.
The synthesis and characterization of fantrip is described in detail in the literature. [10] Likewise, the synthesis of antrip was conducted according to the literature, [11] and the structure of the product was confirmed by 1 H NMR spectroscopy (cf. Fig. S17). 1