Supporting Information to: Impact of device design on the electronic and optoelectronic properties of integrated Ru-terpyridine complexes

The performance of nanoelectronic and molecular electronic devices relies strongly on the employed functional units and their addressability, which is often a matter of appropriate interfaces and device design. Here, we compare two promising designs to build solid-state electronic devices utilizing the same functional unit. Optically addressable Ru-terpyridine complexes were incorporated in supramolecular wires or employed as ligands of gold nanoparticles and contacted by nanoelectrodes. The resulting small-area nanodevices were thoroughly electrically characterized as a function of temperature and light exposure. Differences in the resulting device conductance could be attributed to the device design and the respective transport mechanism, that is, thermally activated hopping conduction in the case of Ru-terpyridine wire devices or sequential tunneling in nanoparticle-based devices. Furthermore, the conductance switching of nanoparticle-based devices upon 530 nm irradiation was attributed to plasmon-induced metal-to-ligand charge transfer in the Ru-terpyridine complexes used as switching ligands. Finally, our results reveal a superior device performance of nanoparticle-based devices compared to molecular wire devices based on Ru-terpyridine complexes as functional units.


XPS measurements
XPS measurements have been performed during the first steps of the Ru(TP)2-complex wire growth to support previously reported IRRAS and Raman spectra [19], since they can provide additional information due to their sensibility towards the oxidation state of ruthenium. Figure S4 shows the C 1s, Ru 3d and O 1s core level spectra. Since the C 1s and Ru 3d core levels, located around 285 eV and 282 eV, respectively, overlap to some extent, we followed the accepted procedure to deconvolute these peaks by fitting the ruthenium contributions first. The Ru 3d peaks are represented by a doublet (3d5/2 and 3d3/2) with a separation of 4.17 eV and an area ratio of 3:2 [28]. As the Ru 3d5/2 peak is well resolved in most cases we will refer to this peak in the following.
Starting the wire growth with the chemisorption of MPTP on the Au substrate the core level spectrum in the range of 278 eV to 290 eV given in Figure S4a displays two C 1s BE. While the peak with low intensity at 286.7 eV (dashed blue line) corresponds to carbon directly bound to nitrogen, the main C 1s peak appears at 285.5 eV (solid blue line) in accordance with literature and contains contributions of all other carbons in MPTP, that is, the residual carbons in the pyridine rings and the phenyl ring (Table 1) [R1]. In sample (ii), the main C 1s peak shifts by 0.5 eV to lower energies as a result of the first wire growth step, the complexation of MPTP with Ru-PF6 in ethanol. This significant effect is caused by the fact that in surface sensitive methods like XPS the contribution of the near-surface cores to the total signal intensity is higher compared to low lying cores. In consequence, the main C 1s peak is shifted to a BE of 285.0 eV due to the contribution of the aliphatic carbons terminating the MPTP-Ru growth step while the contribution of aromatic and pyridine carbons is reduced [R2, R3]. The lower intensity C 1s peak, observed at a BE of 286.2 eV, is assigned to the energy of carbon atoms involved in C-O bonds, like in ethanol [R2]. These C 1s core levels of sample (ii) clearly indicate the presence of ethanol and suggest the formation of a Ru(MPTP)(EO)3-complex (with EO corresponding to an ethoxide anion, CH3CH2O -). The corresponding Ru 3d5/2 peak at 282.1 eV, which can be identified as Ru(III) bound to hydroxide or in this case ethoxide ions, verifies this view (Figure S5b) [28]. In sample (iii), corresponding to MPTP-Ru-BTP wires, and sample (iv), corresponding to MPTP-Ru-BTP-Ru wires, no significant changes of the main C 1s peak can be observed. However, the low intensity C 1s peak shifts alternatingly between 286.5 eV (C-N) and 286.2 eV (C-O) and strongly alternates in intensity, indicating a termination of the Ru-complex wire by either a TP group or a Ru(TP)(EO)3-complex, respectively. A minor C 1s component observed in samples (iii) and (iv) at BE of 288.2 eV and 287.9 eV, is assigned to compounds containing C=O double bonds, like HCO3species, which may result at Au surfaces due to air exposure [R1-R3]. The main Ru 3d5/2 peak observed during the Ru-complex wire growth from sample (ii) to (iv) tends towards a distinctly higher BE (282.1 eV), when the wire is terminated by the Ru (III) (TP)(EO)3-complex (samples (ii) and (iv)) compared to the BE (281.5 eV) attributed to Ru (II) (TP)2-complexes (sample (iii)) ( peak is nearly comparable due to the presence of both Ru-complexes in the wire and the rate of yield of the wire growth steps less than 100% under these mild reaction conditions. An additional Ru 3d5/2 peak with lower intensity at a BE of 280.5 eV is obtained in step (iii). This peak corresponds to hydrated ruthenium oxide, like Ru (IV) O(OH)2 or Ru (IV) O(EO)2, and can be formed due to air exposure. This Ru 3d5/2 peak is also observed in the core level spectrum of sample (iv).
In addition, the O 1s peak assignment reveals the alternation of the Ru-complex wire terminating groups, too. The O 1s core level spectrum of sample (i), MPTP chemisorbed on the Au substrate, shows a residual amount of the solvent ethanol adsorbed on the surface indicated by the O 1s BE of 532.9 eV attributed to aliphatic C-O bonds and 534.2 eV corresponding to carbon bound hydroxy groups involved in hydrogen bonds ( Figure S5b and Table 1) [29,30]. However, the main O 1s peak in the spectra of sample (ii) and sample (iv) appears at BE of 531.8 eV and 531.4 eV, respectively, and indicates the wire termination by the Ru(TP)(EO)3-complex as well as the C 1s and Ru 3d5/2 spectra [29]. On the other side, mainly the signature of ethanol is found in the O 1s core level spectrum of sample (iii) whose termination can be compared with that of sample (i  We assume that transport in a Ru(MPTP)2-AuNP device is based on tunneling of electrons through multiple small tunneling barriers, between an electrode and a AuNP or between two AuNP, respectively, while a bias voltage is applied between the left and the right electrode. The tunneling barriers are formed by Ru(MPTP)2-complexes. Considering the junction geometry given in the schematic, the device conductance (Gdev) is calculated using the series formula

Activation energies of Ru(MPTP)2-AuNP devices
with GL-NP1 and GNP3-R the conductance values resulting from tunneling through the barrier between the left or the right electrode and the nearest AuNP, respectively, while GNP-NP corresponds to the conductance value through the tunneling barrier between two AuNP.
The single-channel Landauer formula is applied to determine the theoretical device conductance at ±1 V according to the method we reported before [15,23,24]. In this model G through Ru(MPTP)2complexes forming the tunneling barrier between an electrode and AuNP or two AuNP is given by: