Molecule‐Electrode Interfaces Controlled by Bulky Long‐Legged Ligands in Organometallic Molecular Wires

Precise control of molecule‐electrode interface is essential for molecular devices. Herein, new ruthenium acetylide molecular wires with long‐legged phosphine ligands to form a sterically controlled molecule‐electrode interface are designed. The sharpened Raman signals ascribed to acetylene stretching are observed for the self‐assembled monolayers (SAMs) of the molecular wires with the biphenyl‐ (2Au) and tert‐butylbiphenyl‐substituted long‐legged dppe‐type ligands (3Au), suggesting that steric hindrance causes formation of uniform SAMs. Scanning tunneling microscope break‐junction (STM‐BJ) study of 3Au reveals narrow conductance features compared with those of 1Au bearing the parent dppe ligands, indicating formation of a uniform molecular junction. Furthermore, the effective electronic interactions between the molecule and electrodes are unique to the long‐legged derivatives, as revealed by the surface‐enhanced Raman scattering study. Thus, the bulky long‐legged strategy turns out to provide a design concept for a well‐defined molecule‐electrode interface.


Introduction
A well-defined, highly conducting molecule-electrode interface is essential for development of organic electronics. Interfacial structures of self-assembled monolayer (SAM) on electrodes www.advmatinterfaces.de and surface-enhanced Raman spectroscopic features when attached to gold surfaces to disclose the steric effect of the ancillary ligands in the molecular wires. The keys of our molecular design are that (i) the charge carriers are transported through the main metallapolyyne-diyl chain, while (ii) the peripheral long-legged supporting ligands indirectly control the interfacial structures between the wires and the metal surface.

Synthesis and Characterization
To form a covalently linked organometallic molecular junction, the terminal carbon atoms were functionalized with the gold fragments. [16] Thus, the gold complexes 2 Au and 3 Au were prepared from the SiMe 3 precursors 2 TMS and 3 TMS , respectively, by treatment with NaOMe and ClAu{P(OMe) 3 } (Figure 1e). Alternatively, 2 Au and 3 Au were be prepared in two steps by the desilylation-auration sequence of n TMS through the terminal acetylene derivatives n H . Their 1 H and 31 P{ 1 H} NMR spectra contain only one set of signals for the ancillary ligands, indicating the longlegged ligands are equivalent, i.e., flexible enough in solution at the timescale of NMR spectroscopy. There is no noticeable deshielding for the TMS or P(OMe) 3 groups, suggesting negligible spatial proximity between the central alkynyl chain and the long-legged biphenyl skeletons. Molecular structures of 2 TMS and 3 TMS are shown in Figure 2a,b. The two alkynyl ligands adopt a trans geometry, and the long-legged ligands are radially spread. The aromatic rings show no π-π interactions (d C-C > 3.5 Å), yet some CH-π interaction is observed for the proximal phenyl rings bonded to the phosphorous atoms. The distances between the terminal carbon atoms of the phenyl rings (≈1.3 nm for 2 TMS ) and the t-Bu groups (≈1.6 nm for 3 TMS ) are longer than those between the terminal carbon atoms (C δ ) of the butadiyne moieties (1.17 for 2 TMS and 1.16 nm for 3 TMS , respectively). Thus, the bulky phosphine ligands may interact with the gold surface when a molecular junction is formed through C-Au covalent bond formation. These structural features resemble those for the optimized structures of 2 H and 3 H obtained by the density functional theory (DFT) calculations ( Figure S22, Supporting Information), being suggestive of negligible packing effect. www.advmatinterfaces.de

SAM Study
To investigate the spatial arrangement of the molecular wires on the gold surface, we carried out Raman spectroscopic study on their SAMs. The gold substrates prepared by frame annealing of gold wires were soaked in 1 mm CH 2 Cl 2 solutions of n Au for 24 h and rinsed with CH 2 Cl 2 . Then, the surface-enhanced Raman scattering (SERS) spectra were recorded and compared with the parent molecules n Au (Figure 3a). The terminal gold functionalization facilitate formation of the covalent Au-C acetylide bonds through either transmetallation or fusion with the gold complexes. [16] The SAM samples are abbreviated as SAMn (n = 1-3). The peaks around 400 cm −1 in the SERS spectra of SAM1-SAM3 may be assigned to the vibrations of the formed covalent Au-C bonds ( Figure S17, Supporting Information). [12] The vibrations around 2050 cm −1 for neat samples of n Au are assignable to the Ru-CC-CC vibrations according to the DFT study ( Figure S19, Supporting Information). On the other hand, SAMn showed properties characteristic of each molecule. Raman signals ascribed to the acetylene parts for SAM1 are observed as ill-defined, broadened signals in the range of 1800-2200 cm −1 , [19] whereas well-defined, sharp signals at 1860, 1898, 1989, and 2040 cm −1 are observed for SAM3. The characteristics of SAM2 are intermediate between SAM1 and SAM3. Because Raman spectra of neat samples of n Au are similar with each other, the differences in SAMn should be caused by different arrangement and connection structures of the molecules on the gold surface.
To consider the interaction modes, we carried out the Raman simulation by the DFT study for the surface-attached models Au-1, where 1 Au is attached to the 17 and 18 gold atom clusters with the on-top, bridge, and hollow modes (Figure 3b). The simulated Raman shifts appear within the range of the experimental ones (1900-2100 cm −1 , Figure 3a). The simulated vibrational peaks result from a combination of two to four acetylene-stretching modes. Peak signatures are sensitive to the surface-bound model and the Au(electrode)-C bond lengths but relatively insensitive to the Au(electrode)-CC angles ( Figure S21, Supporting Information). DFT simula-tion suggests that the broadened signals of SAM1 are caused by the variations of the connection modes and orientations. When compared with the Raman spectrum of SAM3 and the simulated signal pattern and energies, the on-top structure seems plausible among the three fashions. The on-top structure should be induced by the long-legged ligands to avoid steric repulsion between the bulky ligands and the electrodes. Molecular modeling of 3 H on a flat Au surface suggests that the phosphine ligand is flexible and that 3 H can stand on the surface in a way that avoids steric hindrance to the Au surface ( Figure S24, Supporting Information). SAM2 showed the spectrum similar to but rather broader than that of SAM3. This result suggests that the medium-sized biphenyl ligands are also effective to control the binding modes, while biphenyl derivatives have some degree of structural freedom on the Au surfaces compared with the tert-butyl-biphenyl derivatives.

STM-BJ Study
We performed single-molecule conductance measurements of the molecular junctions of Au-2-Au and Au-3-Au prepared from 2 Au and 3 Au , respectively, by the STM-BJ method, [20] and compared with the conductance of Au-1-Au. As we previously reported, 1 Au in-situ formed covalently linked molecular junction Au-1-Au, and the single-molecule conductance of the junction was 2.1 × 10 −2 G 0 (log G M /G 0 = -1.77). The samples were dissolved in tetraglyme (0.25 mm) and bias voltage was set to 100 mV. Typical conductance traces and histograms for Au-n-Au are shown in Figure 4. STM break-junction measurements of Au-2-Au and Au-3-Au also revealed steps and peaks in the individual traces and log histograms, respectively. These results indicate that molecular junctions Au-2-Au and Au-3-Au formed despite their bulky long-legged ligands. Based on deconvolution analysis, the most probable single-molecule conductance (log G M /G 0 ) is determined to be -1.81 (2 Au ) and -1.71 (3 Au ), which are virtually identical to that of 1 Au . Thus, it can be concluded that the carrier transport occurred through the Ru(CC-CC-C) 2 chain regardless of the phosphine ligands. Furthermore, their full widths at the half maximum (FWHM) of the log-scale conductance histograms (Figure 4a and Table 1) were sensitive to the phosphine ligands. As the length and bulkiness of the phosphine ligands increase, the FWHMs significantly decrease (1.16 (1) → 0.72 (2)→ 0.32 (3)). The individual traces (Figure 4b) reflect the trends observed for the log histograms. In contrast with the relatively flat steps observed for 3, the steps for 1 were fluctuated. Thus, the long-legged ligands turned out to be able to control the conductance distribution.
The most probable binding mode of Au-1-Au is presumed to be the on-top structure as reported previously, [16] but we do not exclude the possibility of other binding modes in the BJ The arrows indicate conductance peaks. The conductance around log G/G 0 = -0.5 is caused by the scattered conductance of Au nanowires. [23,24]

www.advmatinterfaces.de
process. The binding modes occurring in diethynylbenzene-diyl junctions are known to impact conductance, [21] and, thus, the presence of a range of binding modes for the acetylide junctions would also result in variations in conductance. As discussed in the SAM study, binding modes on the Au surface can be controlled by the long-legged ligands, which would also effective to control the binding modes in the SMJ and contribute to the narrow conductance distribution.
Furthermore, we are also interested in the effects of molecular orientation on conductance, which have not been examined in the literature. To gain insight into this point, we carried out the DFT-non equilibrium Green's function (NEGF) calculation. Thus, the plots of the tilt angles ∠CC-Au (θ) versus conductance are shown in Figure 5, using the molecular junction models with the Au 35 clusters. As the angle θ decreases, transmission at the Fermi level is raised by more than one order of magnitude at 140°. Because the main conduction orbitals are not dependent on θ, the enhanced conductance with an increase of θ is probably caused by the changes in the degree of the dπ-pπ orbital interactions of the Au-C bonds, which was previously proposed to affect conductance. [11] In this context, the bulky long-legged ligand may hinder the formation of the molecular junction with smaller θ values by steric repulsion, leading to suppression of the variation of conductance in the molecular junction. [22] The most probable conductance in Au-n-Au is similar regardless of the ligands, suggesting that the mole-cules are fully stretched with θ close to 180°. Therefore, both the restricted angles and binding modes of the long-legged ligands should contribute to the narrow conductance distributions.

SERS-MCBJ Study
We carried out SERS measurements during the mechanically controllable BJ (MCBJ) processes to obtain further information on the electronic structures of the molecular junctions. [25,26] We recorded Raman spectra for Au-1-Au and Au-3-Au when the junction conductance reached around 10 −2 G 0 , suggesting molecular junction with the on-top attachment. Both Au-1-Au and Au-3-Au junctions showed Raman signals around 400 cm −1 assignable to the Au-C covalent bonds ( Figure S18, Supporting Information). [12] In contrast with SAM1, the SERS spectrum of the Au-1-Au junction showed the simple spectral features with the major peak at 2025 cm −1 accompanying the satellite peak at 1998 cm −1 (Figure 6), and both are ascribed to the ν(CC) vibrations. On the other hand, the Raman spectrum of Au-3-Au junction with the bulky substituents showed the single and sharp Raman peak at 1972 cm −1 .
The Raman spectra contain information on dynamic BJ processes. Though we fixed the junction structure, the motion of the molecule and gold atoms modulate the junction structure. Therefore, the rather broadened and multiple

Conclusion
In summary, we have developed organometallic molecular wires with the long-legged ligands (2 R and 3 R ), where the metal-molecule interfacial structures are controlled by the ligands. The bulky long-legged ligands provide self-assembled monolayers with sharp SERS CC vibration signals, suggesting formation of well-defined SAMs. Furthermore, STM break-junction study reveals that the long-legged ligands make the conductance distribution narrower. SERS-MCBJ study supports controlled interfacial structures by the long-legged ligands at the single-molecule junction. Thus, our study demonstrates the successful indirect control of molecule-electrode interfacial structures by long-legged ligands. This long-legged strategy will be useful for applications toward various molecular devices.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.