N‐Heterocyclic Carbenes: Molecular Porters of Surface Mounted Ru‐Porphyrins

Abstract Ru‐porphyrins act as convenient pedestals for the assembly of N‐heterocyclic carbenes (NHCs) on solid surfaces. Upon deposition of a simple NHC ligand on a close packed Ru‐porphyrin monolayer, an extraordinary phenomenon can be observed: Ru‐porphyrin molecules are transferred from the silver surface to the next molecular layer. We have investigated the structural features and dynamics of this portering process and analysed the associated binding strengths and work function changes. A rearrangement of the molecular layer is induced by the NHC uptake: the NHC selective binding to the Ru causes the ejection of whole porphyrin molecules from the molecular layer on silver to the layer on top. This reorganisation can be reversed by thermally induced desorption of the NHC ligand. We anticipate that the understanding of such mass transport processes will have crucial implications for the functionalisation of surfaces with carbenes.


Sample preparation
Three different ultra-high vacuum setups were used to perform the measurements. In all chambers, similar procedures were employed to prepare the samples. The Ag(111) single crystal went through several cycles of Ar + or Ne + sputtering and subsequent annealing to 725 K.
Ru-TPP was deposited onto the Ag(111) crystal, which was held at 300 K, by organic molecular beam epitaxy (OMBE) of Ru(CO)-TPP (Sigma Aldrich, 80 % dye content) at 550 K to 625 K. It has been shown, that clean Ru-TPP monolayers are obtained as the CO detaches from the Ru-TPP in this process and the purity of the compound can be enhanced by thorough outgassing. 1 The deposition rates were 0.3 molecules nm 2 h to 2.0 molecules nm 2 h , depending on the temperature and chamber geometry. Monolayers of Ru-TPP self-assembled in the compressed phase were obtained by multilayer desorption at 550 K.
The masked IMe (1,3-dimethyl-1H-imidazol-3-ium-2-carboxylate,IMe-CO 2 ) was synthesized according to a reported procedure. 2 The free IMe was similarly deposited via OMBE by heating the masked IMe to temperatures of 360 K to 390 K. Deposition rates of up to 10 molecules nm 2 h were observed on Ru-TPP monolayer samples held at 300 K, whereas at 200 K the rates were ∼5 times higher.

STM
An Aarhus-type STM (SPECS GmbH) was used. The STM was operated in constant current mode with a chemically etched W tip, the tunnelling bias was applied to the sample.

XPS
XP spectra were acquired with a Mg Kα source and a SPECS Phoibos 100 CCD hemispherical analyser in normal emission geometry. The energy scale was calibrated using the Ag 3d 5/2 peak at a binding energy of 368.27 eV.

Work function determination
The work function was determined by the detection of the secondary electron cutoff by using the following relationship: where Φ is the work function, hν is the excitation photon energy, E Fermi level is the energy of the Fermi edge and E secondary electron cutoff is the energy of the secondary electron cutoff. For the measurements of the secondary electron cutoff, the sample was set to a potential of -20 V with respect to the electron analyser. The electrons were excited with Al Kα (hν = 1486.61 eV) source and detected by a SPECS Phoibos 100 CCD hemispherical analyser in normal emission geometry. The energy scale was calibrated by the Ag 3d 5/2 peak at a binding energy of 368.27 eV. The data collected for the work function determination are presented in the following figure against a calibrated energy scale. The secondary electron cutoff (marked by a dashed line for the data corresponding to the clean Ag(111) surface) in kinetic energy scale amounts to the work function (Φ Ag(111) ). The data presented in manuscript Figure  4 are a zoom in the secondary electron cutoff region vs. kinetic energy.

LEED
LEED measurements were performed with a BDL800IR-LMX-ISH by OCI Vacuum Microengineering Inc. All images were acquired with an electron energy E = 20 eV. Subtle changes in spot size of the same LEED pattern are associated with slightly worse molecular ordering or presence of spurious defects, e.g., reduced extension of the crystalline domain size.

TPD
Measurements were performed using a quadrupole mass spectrometer behind a copper cap cooled with LN 2 , 3,4 with the sample located at a distance of ∼ 1 mm from the opening on the apex of the copper cap. All here reported spectra show the parent ion of IMe with m/z = 96. The heating rate for all spectra was set to β = 5 K s −1 .

TPD Analysis
Assuming that the desorption curves can be described by the Polanyi-Wigner equation, the peak temperature with increasing initial coverage θ 0 can be modelled by a coverage dependent desorption energy E des . The inclusion of a decreasing desorption energy at increasing coverage θ can be used as a model for repulsive interactions. 5, 6 We employ a linear decrease of the desorption energy with coverage, leading to the following equation: This model replicates the shape of the TPD spectra as shown in Figure S13. NIXSW X-ray standing wave profiles were acquired at the I09 beam line at the Diamond Light Source. 7 All measurements were acquired with the sample held at ∼200 K, using a Scienta EW4000 HAXPES analyser that was mounted perpendicular to the incident X-rays in the horizontal plane of the photon linear polarisation. Measurements for the (111) Bragg reflection with Bragg diffraction planes parallel to the surface were performed at a normal incidence Bragg energy of hν = 2.63 keV. All measurements were repeated multiple times at different spots of the sample, where at each spot the reflectivity curve was measured to allow a precise energy alignment of the individual NIXSW measurements and to ensure the crystalline quality of the Ag(111). Monitoring of potential beam damage was performed by recording XP spectra of the C 1s and Ru 3d region before and after each NIXSW measurement.

DFT
The DFT geometry optimisation was carried out via the Quantum ESPRESSO 8 package. Within the vdW-DF2-B86r approximation 9 in the exchange-correlation term, five layers of the Ag(111) substrate were considered, the two lower layers fixed at their bulk-terminated positions. An optimized lattice constant of 4.1325 Å, 2×2 k points, Fermi-Dirac smearing of occupation numbers with a 50 meV broadening, projector augmented wave (PAW) 10 data sets for the pseudization of the core electrons, surfacedipole corrections, and cutoff energies of 60 Ry for the wave functions and 350 Ry for the electron density were applied. The unit cell included two molecules, as derived for square phase Ru-TPP by STM and LEED. 1

Dipole strength estimation
The reduction of the adsorption energy due to dipole interactions is estimated in a simple model on the basis of Coulomb-interactions between dipoles. The potential energy between two point charges q 1 and q 2 with a separation r is given by

S-5
To extend this to dipoles, the two charges of each dipole, separated by distance l, have to be considered. A superposition of energy contributions from surrounding dipoles, assuming a full coverage of IMe, leads to the following equation.
Including now as a last step image dipoles, induced by the conductive Ag(111) substrate, we obtain Here s is the distance between a dipole and its image dipole. With estimates for the required charges and lengths (q = 0.3 e, obtained from DFT as charge on the IMe; l = 4 Å as an estimation for the separation of IMe and the TPP macrocycle; s = 6 Å as an approximation of two times the Ag(111)/Ru-TPP separation), which results in a dipole moment of p = q · l = 6 D, E Dip = 0.12 eV is obtained, assuming a square assembly of the Ru(IMe)-TPP molecules with a spacing of 1.3 nm, determined for the compressed phase. The energy difference between calculations with total numbers N of surrounding dipoles of 10000 and 40000 was less than 1 %.
Scheme of the dipole-dipole interaction. The two dipoles (black ellipses) are separated by distance d, the charges in each dipole by l, the distance between dipole and image dipole (grey) is s.

S-6
Additional Experimental Data Figure 11 The shift in energy is consistent with a decoupling of the Ru-TPP from the Ag(111) surface by the introduction of an axial ligand, firstly reported for Co-TPP with NO. 12 For the particular metalloporphyrin investigated here (Ru-TPP) on Ag(111), a shift to 281.8 eV is found after ligation of CO. 3 The slightly lower energy shift induced by IMe can be attributed to the electron donating character of the ligand, evidenced further in the DFT and TPD analyses as well as the work function measurements, presented in the manuscript. Additionally, an increase of the Ru 3d 5/2 signal can be observed. This is similar to the XPS changes observed after CO axial ligation on the Ru atoms of Ru-TPP/Ag(111) 13 and is tentatively attributed to the change of the Ru environment, promoting forward scattering of the emitted photoelectrons. This effect is enhanced at high kinetic energy, due to the specific angular dependence of the atomic scattering factor for heavy metals. 14 Another probable, concomitant cause is the change of the coordination of the Ru upon ligation, which alters the crystal splitting of its d-band. Combined with decoupling the Ru from the metallic substrate (and thus the substrate's delocalized electrons), it could substantially alter the loss structure of the core level, resulting in the apparent increase in intensity in those spectra. The C 1s XPS signal changes in shape, and its total area increases by 5%. A shape change is expected due to the contribution from the C 1s of the IMe ligand at slightly higher binding energies, 2 resulting in the shoulder visible to the left of the main peak as well as the Ru 3d 3/2 contribution shifting from 283.6 eV to 284.8 eV. We further note that the rearrangement of the surface might cause shadowing, which will affect the related peak intensities. No noticeable screening effect is expected for Ru(IMe)-TPP on Ag(111) based on earlier studies of Ru(CO)-TPP on the same surface. 13 Figure S2: Fitted spectra of the Ru 3d 5/2 signal resulting from the sum of all spectra used for the NIXSW analysis. Two different components can be clearly distinguished, Ru-TPP (orange fit) and Ru(IMe)-TPP (blue fit). The fitting parameters derived from this analysis were used for the analysis of spectra recorded at different photon energies, shown in Figure S3.
S-8 Figure S3: Fits for the Ru 3d 5/2 spectra (intensity in arbitrary units vs. binding energy in eV) of the NIXSW analysis presented in the manuscript Figure 1c        S-18