Aurophilic Molecules on Surfaces. Part II. (NapNC)AuCl on Au(111)

Although aurophilicity is a well-known phenomenon in structural gold chemistry and is found in many crystals of Au(I) complexes, its potential for self-assembly in thin films is not yet explored. This paper is Part II of a study, in which we investigated the ultrathin film formation of chlorido(2-naphthyl isonitrile) gold(I) on gold surfaces. Here, we present the data for the growth of (NapNC)AuCl on isotropic Au(111) surfaces. Already during physical vapor deposition, the condensation of ultrathin films is monitored by photoelectron emission microscopy (PEEM) and incremental and spectrally resolved changes in the optical reflectance (DDRS). Additional structural data obtained by scanning tunneling microscopy (STM) and low-energy electron diffraction (LEED) reveal that the “crossed swords” packing motif known from the bulk is also present in thin films.

1 Dierential Reectance Spectroscopy Figure 4 of the main article shows the incremental changes of the reectivity between 2 eV and 4 eV.Similar results were obtained for the deposition of (NapNC)AuCl on Au(110) surfaces (see ref 1).Here we want to rationalize, at least qualitatively, the amplitude and sign of the observed spectral changes.
To this end, we have modeled the optical reectance for a so-called three-layer-system consisting of (1) vacuum with refractive index N 1 = n 1 = 1, (2) an organic thin lm with anisotropic refractive index N 2 = (N 2x , N 2y , N 2z ) and thickness d, and (3) a thick gold substrate with complex refractive index N 3 = n 3 + ık 3 as tabulated in ref 2. A sketch of the model is shown in the insets of Figures S1 and S2.As reported by Hobbollahi and coworkers in ref 3, the optical transitions of (NapNC)AuCl are located in the UV range (E ≥ 4 eV, λ ≤ 300 nm).Away from the absorption resonances, the refractive index is essentially real valued (N 2i = n 2i + ık 2 ≈ n 2i with i ∈ {x, y, z}) and its wavelength dependence can be described by a Cauchy term: or (in a rst approximation over a limited spectral range) simply by a constant value n 2i ≥ 1.
In fact, ref 4 reports optical data for common organic molecules containing naphthyl groups.The complex refractive index for these molecules are almost featureless in the visible range, besides a steady, but small increase of n and k towards the UV.Moreover, n is about a factor 30 larger than k.Therefore, our subsequent assumption of constant, real-valued components for the (anisotropic) refractive index N 2 with values n 2i in the interval between 1 and 2 seems well justied.
According to Azzam (see ref 5 and there in particular chapter 4.7.3.4), for the case of a biaxially anisotropic lm (2) on an isotropic substrate (3) in an isotropic ambient medium (1), where the principal axes x and y of the biaxial lm lie in the lm plane, while the third axis (z) is perpendicular to the lm plane (i.e., parallel to the surface normal) the reection coecients for s-or p-polarized light for the entire three-layer-system are as follows: S2) If the scattering plane is located in the (x, z)-plane, then the so-called phase thicknesses β s and β p are dened as follows: (S4) The Fresnel equations give the reection coecients r 12s and r 12p at the interface between layers 1 and 2 In the above equations, θ 1 denotes the angle between the light beam in the ambient medium (1) and the direction normal to its interface with layer 2. The equations above can also be used to calculate the reection coecients r 23s = −r 32s and r 23p = −r 32p between layers 2 and 3 by replacing the index 1 by 3 and using Snell's law N 3 sin θ 3 = N 1 sin θ 1 .Likewise, the reection coecients from the bare surface r 13s , and r 13p can be obtained using the same formula.In the latter case, the Fresnel equations simplify to the well known form for the interface between two optically isotropic media.
Note that on the Au(111) substrate with an in-plane three-fold rotational symmetry, the (NapNC)AuCl overlayer will have a net laterally isotropic refractive index, i.e., N 2x = N 2y , whereas on the Au(110) substrate the molecules in the overlayer are uniaxially aligned 1 and In a next step, the reectances for p-and s-polarized light are obtained from the corre-  S2, respectively.For N 3 , we used a linear interpolation of the data-set of Johnson and Christy for polycrystalline gold published in the refractive.indexdatabase. 6Since the angle of incidence, θ 1 = 65 • , is close to the Brewster angle for gold, the reectance for s-polarized light is signicantly higher than for p-polarized light, especially above ∼ 2.5 eV.
To explore the eect of the (NapNC)AuCl overlayer, we calculated the DRS spectra Case B is certainly the simpler one: The s-polarized light is not aected by an out-of-plane component of the refractive index since only the p-polarized light probes the n 2z component.
The corresponding results of the simulation for case B are shown in Figures S1c and S2c.
As expected, the DRS signal for s-polarized light is a at line equal to 0. The p-polarized light corresponds to an increased reectivity of the sample.Although the assumed (real) refractive index of the organic layer shows no dispersion, i.e., is constant across the entire photon energy range shown, the DRS signal changes most, where the reectance of the bare sample is small.Therefore, the resulting DRS signal resembles the shape of the reectance of the bare gold surface (see Figures S1a and S2a).A refractive index with mainly an outof-plane component should be characteristic for (NapNC)AuCl dimers on the surface.In such a conguration, the molecules are not only tilted with respect to the surface but also slightly bent.The resulting dipole layer might thus also aect the charge distribution of the topmost layer of the substrate.
Case A represents (NapNC)AuCl molecules with only an in-plane component of the refractive index.In addition, the refractive index is assumed to be isotropic in the (x, y)plane.As mentioned earlier, this is certainly valid for molecules adsorbed on the optically isotropic Au(111) surface.The calculated DRS spectra reveal a decrease of the reectance for both p-and s-polarized light, as shown in Figures S1b and S2b.Note that the amplitude of the change in reectivity for p-polarized light is about a factor of four larger than that for s-polarized light.This results from the angle of incidence being close to the Brewster angle.Such a dierence in the signal amplitudes was also observed in the experiment (see Figures 4c and 4d of the main article).As in case B, the spectral lineshape follows that of the reectance of the bare surface.However, the reectance is now decreasing and not increasing as in case B. Since we can reproduce the correct order of magnitude of the changes as well as their sign, the assumed combinations of layer thickness and refractive index of the organic layer seem to describe the real system rather well.
2 Temperature and Voltage Dependent STM Figure S3 shows STM images taken after deposition of Θ ≈ 1.2 ML (NapNC)AuCl on a Au(111) single crystal held at room temperature.In both cases, the deposition was stopped immediately after the plateau of the PEEM transient (see Figure 4a of the main article).The surface shown in Figure S3a was imaged without further heat treatment at room temperature: two domains are visible.The stacking direction of the dimers along the short unit cell axis diers bei about 22 • between the two domains.This indicates that the imaged surface is covered by the M 1 structure as discussed in the main article.It also shows that the domains extend over more than 100 nm in each of the two dimensions on the surface.The noise in the image indicates that molecules are present in the second layer forming a 2D gas phase.
Such movement of molecules in a 2D molecular gas phase can be prevented if the sample is cooled during STM image acquisition.Therefore, the sample shown in Figure S3b was cooled with liquid nitrogen to about 110 K.The sample is not identical to that shown in Figure S3a, but was prepared in an identical process and with a similar nal coverage of about 1.25 ML.We expect that about a quarter of the surface area is already covered with molecules in the second layer.Therefore, we interpret the lower-appearing, dark structures in the image as condensed molecules in the second layer.As expected, the darker areas cover roughly 25 % of the imaged area.Thus, we can conclude that a sample temperature of 110 K is sucient to freeze out the diusive motion of the molecules in the second layer.not move.We can assume that the structure shown belongs to the rst layer.For other parameters (not shown here), we were able to tunnel directly into the gold atoms of the substrate.We also observe that the same area shows an increasing defect density after repeated scanning with the STM tip.Due to such tip induced changes, the parameter range and time window allowing to resolve the dimer structure as in Figures 3a and 4a of the main article is rather limited.

Position Dependent LEED Images
Figure S4 shows a series of LEED images taken while manually (and therefore randomly) varying the position of the sample with respect to the impinging electron beam.The LEED patterns were recorded with an electron energy of 22.8 eV.To image the spots close to the central (0,0) reex, the Au(111) single crystal was tilted slightly with respect to the LEED optics.The deposition of (NapNC)AuCl was stopped upon reaching a coverage Θ of about 1.2 ML: after a plateau, the transient of the mean electron yield decreases steeply here.
Before the sample was examined in the LEED, it was annealed at 353 K for a total time of about 20 min.A comparison of Figure S4 with Figure 5a from the main article conrms that the surface is mainly covered with the molecular phase, which can be described by the following epitaxial matrix: Due to the symmetry of the substrate, we can assume that six equivalent domains contribute to the LEED image.When the position of the sample in front of the LEED optics was changed, certain groups of spot lose intensity while others gain intensity.These intensity variations were used to identify spots belonging to one of the six equivalent domains.
Since at some sample positions only one or two (instead of all six) orientations of the dense phase contribute to the diraction pattern, it is tempting to assume that the sample area hit by the electron beam is covered with only one or two domains (see eq S8).This might indicate that extremely large domains formed during the annealing process, e.g. by Ostwald ripening. 7Since terraces of the single-crystalline Au(111) substrate usually extend over a length of about 500 nm, such uniformly ordered domains should extend across many terrace boundaries.A more likely explanation could be that locally higher step densities of the substrate can trigger the molecules to preferentially arrange in a particular orientation with respect to these step edges.
A striking feature in all LEED images in Figure S4 is the bright halo around the (0,0) spot.
The radius of the halo corresponds to the approximate size of the (NapNC)AuCl dimers.
Since the total amount of molecules deposited lls more than a densely packed layer, it is likely that this halo corresponds the incoherent superposition of structure factors originating from the diraction of individual dimers.These dimers form a dilute 2D gas phase in the second layer allowing for a certain rotational degree of freedom in the 2D plane. 8,9

FigureFigure
Figure S1: (a) simulated reectance of the bare Au(111) surface for p-polarized light incident on the surface at an angle of 65 • to the surface normal.Simulated DRS signals for p-polarized light caused by an additional anisotropic layer (2) with varying in-plane (b) and out-of-plane (c) components of the refractive index n 2 .The simulation was carried out for two dierent thicknesses d of the intermediate layer.
Figures S1 and S2, respectively.For N 3 , we used a linear interpolation of the data-set of John- Figures S1 and S2, the changes in the dierential reectivity for both p-and s-polarized light scale linearly with the thickness d (as expected for d ≪ λ).For our simulation, we considered two basic cases: (A) the thin layer has only an in-plane isotropic contribution, i.e., the z component of the refractive index, n 2z is set to 1, versus (B) the thin layer has only an out-of-plance contribution, i.e., the in-plane components While no molecular structure is visible in the bright and dark regions of the large area scan, only a slight change in the tunneling parameters reveals at least the existence of rows similar to the ones observed in FigureS3a.The defects imaged in the inset of Fig S3bdo

Figure S3 :
Figure S3: STM images of ≈1.25 ML (NapNC)AuCl deposited on a Au(111) surface.Both images show an area of 100 × 100 nm 2 .(a) During imaging the sample was at room temperature.The STM parameters were U sample = −1.0V and I T = 7 nA.A domain boundary, which appears particularly noisy, can be seen in the image.(b) The image was acquired at a sample temperature of about 110 K with U sample = 700 mV and I T = 15 nA.The inset shows a 20 × 20 nm 2 large area on the surface after changing the STM parameters to U sample = 1.0 mV and I T = 25 nA and relocation of the tip.The solid white lines indicate the stacking direction of the dimers along the short axis of the unit cell.

Figure S4 :
Figure S4: Series of LEED images taken while the Au(111) single crystal was randomly moved in front of the LEED optics.Therefore, dierent areas of the sample some 100 µm apart are imaged in panels (aj).The (NapNC)AuCl coverage is Θ ≈ 1.2 ML.The sample was annealed for a total of 20 min at a temperature of 363 K.The energy of the electron beam was 28.2 eV.