Redox-Active Ferrocene grafted on H-Terminated Si(111): Electrochemical Characterization of the Charge Transport Mechanism and Dynamics

Electroactive self-assembled monolayers (SAMs) bearing a ferrocene (Fc) redox couple were chemically assembled on H-terminated semiconducting degenerate-doped n-type Si(111) substrate. This allows to create a Si(111)|organic-spacer|Fc hybrid interface, where the ferrocene moiety is covalently immobilized on the silicon, via two alkyl molecular spacers of different length. Organic monolayer formation was probed by Laser Ablation-Inductively Coupled Plasma-Mass Spectrometry (LA-ICP-MS) and X-ray photoelectron spectroscopy (XPS) measurements, which were also used to estimate thickness and surface assembled monolayer (SAM) surface coverage. Atomic force microscopy (AFM) measurements allowed to ascertain surface morphology and roughness. The single electron transfer process, between the ferrocene redox probe and the Si electrode surface, was probed by cyclic voltammetry (CV) measurements. CVs recorded at different scan rates, in the 10 to 500 mV s−1 range, allowed to determine peak-to-peak separation, half-wave potential, and charge-transfer rate constant (KET). The experimental findings suggest that the electron transfer is a one electron quasi-reversible process. The present demonstration of surface engineering of functional redox-active organometallic molecule can be efficient in the field of molecular electronics, surface-base redox chemistry, opto-electronic applications.

kind of supramolecular architectures were experimentally characterized in an extensive way as far as concerns the electron transfer ET process 19,21,[23][24][25][26][27] , while from a theoretical point of view the fundamental knowledge of the ET dynamics, at an hybrid interface, is a subject yet open for discussion 20,28,29 . Indeed, most of the studies so far refer to ferrocene-terminated alkane thiols attached to a gold substrate. For example, a recent study by Wong et al. show that Fc-alkanethiol SAMs form strong ion pairs in the oxidized state which increase the chain length by more than 1 nm as probed by combining electrochemical and XPS measurements 30 . However, techniques to assemble the ferrocene derivative on the different substrates may vary depending on the functional group available. Thus, the functionalization of a molecule-electrode for studying interfacial properties needs a special attention. Different fruitful methodologies exist for the creation of hybrid interfaces, relying on the covalent grafting of functional organic molecules, organometallics, inorganic complexes and these are generally based on ultra-high vacuum (UHV) depositions [31][32][33][34] , wet chemistry 1,19 , electrochemical based methodologies 35 . Fc-SAMs other than the thiolated groups, especially Fc containing alcohol groups are basically unexplored. In this paper, we focus on silicon-based interfaces, as silicon is a linchpin material for micro-electronic industries and can be chemically functionalized by covalent grafting of the ferrocene derivatives. The wet chemistry method is the most widely used technique, due to the simple lab procedures and the vast number of reactive groups that can be grafted on the varied substrates. Herein, we study laboratory synthesized ferrocene derivatives covalently grafted on freshly prepared H-terminated silicon (111) substrate (Fig. 1).
This study is carried out by varying the length of the chain (spacer) linking the ferrocene moiety to the surface: (1) methanol-ferrocene directly grafted on the surface (2) ferrocene grafted through a 1-iodoundecanoic acid (UA) monolayer, Fig. 2.
Formation, surface coverage, elemental analysis, electron-transfer kinetics of the Fc-SAMs on H-terminated silicon was ensured by X-ray Photoelectron Spectroscopy (XPS), Laser Ablation-Inductively Coupled Plasma-Mass Spectrometry (LA-ICP-MS), cyclic voltammetry (CV) measurements. The experimental outcome is compared to theoretical results concerning the dynamics of the electron transfer as obtained by application of the Marcus theory 36 . From the theoretical point of view, our main goal is to verify if Marcus theory 37 is correctly capable in reproducing ET rate constants obtained experimentally. Note that, some recent works 20,28 were devoted to point out that the Marcus model, which describes the ET dynamics in the incoherent limit 38 , fails to give a correct prediction of the ET rate constants. Particular attention is also devoted to draw a parallelism between the striking differences observed for the ET dynamics, probed by CV measurements, in the case of ferrocene covalently immobilized, or in bulk solution, in the case of SAMs of comparable thickness.  [39][40][41] . Experimental data show a spin orbit splitting at 13.2 eV and a similar spin-orbit splitting was reported with other Fe II -polypyridyl complexes 39 . The broad shoulder at about 711 eV is Step (b) covalent grafting of hydroxymethyl ferrocene on H-terminated silicon. www.nature.com/scientificreports www.nature.com/scientificreports/ instead related to the presence of some amount of ferrocenium, which has been typically observed for these systems, either due to the manipulation of the samples or to the irradiation of the films under the X-ray beam [39][40][41] . A shake-up satellite contributes to the high binding energy tail of Fe 2p 3/2 , at about 714 eV [39][40][41] . In Fig. 3b the Si 2p is shown. The pristine surface presents a single peaked structure centred at 99 eV, indicating no traces of surface oxidation 42,43 . Indeed, a non-negligible amount of oxidised Si is found on the Si−UA−Fc surface. To assess quantitatively the Si(111) surface coverage, the attenuation of the Si 2p signal from the substrate was considered, assuming an inelastic mean free path (IMFP), λ, of 31 Å (at the kinetic energy of the Si photoelectrons of 1150 eV) in the organic layer 44 . The film effective thickness was derived applying the standard formula λ θ = ( )

Results and Discussion
1° is the average emission angle and I Si p d 2 ,0( ) are the peak intensities of the Si2p levels from the bare substrate and from the covered sample with a film of effective thickness d, respectively 45 . We obtained an effective thickness, d = 32 Å (with a tilt angle of about 80 degrees with respect to the silicon surface) and 16 Å (with a tilt angle of about 68 degrees with respect to the silicon surface) for the Si−Me−Fc and Si−UA−Fc surfaces, respectively. This result may suggest, by comparison with optimized structures and relevant geometrical parameters obtained by DFT results (shown in Fig. 1), that the coverage with Si−Me−Fc is about 2 to 3 monolayers, while the coverage corresponds to a single monolayer in the case of Si−UA−Fc. This also accounts for the reduced intensity of Fe 2p structures in Si−UA−Fc compared to Si−Me−Fc.
Electrochemical studies. Attachment of the redox-active ferrocene moiety and its electrochemical kinetic studies were performed by conventional CV measurements. Figure 4a,b shows CV curves as a function of the scan rate, the latter are recorded by using the Si−Me−Fc and Si−UA−Fc interfaces as the working electrode. In the case of the Si−Me−Fc a one electron oxidation process was observed at +0.45 V, while reduction process occurs at +0.35 V (vs. SCE), compare Fig. 4a. As expected the oxidation process shifted towards more positive potential, while reduction process shifted towards more negative potential upon increasing the scan rates. Indeed, the peak-to-peak CV separation (ΔE p ) increases with the scan rate (compare Fig. 3SI in the Supporting  www.nature.com/scientificreports www.nature.com/scientificreports/ Information). From a qualitative point of view, the cyclic voltammetry results suggest a quasi-reversible in nature oxidation process of the grafted ferrocene, this because the peak potentials of the forward and backward scan are always shifted by a not-negligible voltage value, while a true reversible process should feature the same (reduction and oxidation) potential values 3 . Using the cyclic voltammetry data, surface coverage of the Si−Me−Fc monolayer was determined at 4.3 × 10 −10 mol/cm 2 considering i p vs. ln(ν) 46 . For a short molecule, this surface coverage value indicate that monolayers are reasonably well packed [47][48][49] . Linear behaviour of the Faradic peak current as a function of the scan rates strongly indicate that Fc−CH 2 −OH successfully covalently grafted on the silicon substrate. The ratio between anodic and cathodic peak current was experimentally found just larger than one (again in agreement with a quasi-reversible behaviour). Figure 4b sets out CVs recorded in the case of the Si−UA−Fc electrode, the overall behaviour is similar to that one of the shorter molecular-spacer with a more prominent less-reversible behaviour: larger ΔE p values and smaller current values with respect to the Si−Me−Fc system. We further investigated full width at half maximum (FWHM) at different scan rates, which were found higher than the ideal Nernstian behaviour of 90.6/n mV. This deviation can be ascribed to stray electrostatic effects induced by adjacent charged species not involved in the redox process 50 . Table 1 sets out the experimental parameters obtained by treatment of data in the Fig. 4a,b. A substantial large variation is observed in the electron-transfer rate constant, K ET , experimental values present in the literature relevant to ferrocene moieties grafted on Si single crystal surfaces (ranging between 4500 to 3 s −1 , such a large variation indicates a high sensitivity of electrochemical measurements to the structure, which is a function of the chemical preparation, of these hybrid interfaces) 25,43,44,[47][48][49]51 . Note that all the elaboration of the experimental electrochemical evidence (details are reported in the Supporting Information) was carried out assuming a one electron single step charge transfer mechanism, allowing for an effective rationalization of the experimental overall picture. Which is the same mechanism underlying the theoretical analysis reported in ref. 34 Table 1) ii) the peak-to-peak potential separation (ΔEp) is smaller for Si−Me−Fc than for the Si−UA−Fc: for the same value of scan rate, compare for instance the potential peak-to-peak separation in Fig. 4a  Atomic force microscopy. To assess the influence of the functionalization process on the morphology of the surface, AFM images were recorded. Figure 5a shows the surface morphology of the Si−UA−Fc hybrid interface, which is the sample subjected to the largest number of electrochemical and chemical treatment steps.  www.nature.com/scientificreports www.nature.com/scientificreports/ bonds (even in the case of a long, in principle not conducting, alkyl chain) acts as an on/off switch on the ET process involving the ferrocene: the charge of the interface is driven by the external applied potential (bias), which can be switched on (charged, i.e. oxidized ferrocene) and off (discharged, i.e. neutral ferrocene) following the application of a suitable value of the bias. Moreover, the results here reported show that the efficiency of the on/ off (charge/discharge) process is a function of the structure of the linker placed between the redox centre (ferrocene) and the electrode surface (in this case silicon). This result is at variance with respect to the case of an alkyl SAM functionalized electrode surface (1decanethiol on Au), but featuring the ferrocene in bulk solution: in this case the Faradaic current is almost zero although the distance between the ferrocene and the electrode surface, as well as the chemical nature of the spacer, are quite similar 4 . Moreover, a semi-quantitative agreement between the experimental and theoretical values is found for the "short spacer" ferrocene derivative, also XPS data reveals well-defined signals including elemental compositions. For the Si−UA−Fc system a much larger, than theoretically expected, experimental K ET value is found. A result in line with findings recently discussed in the literature, where high current values are found (i.e. CVs featuring evident redox current peaks) even dealing with long saturated alkyl-carbon chains, where a non-conducting behaviour is expected 3,20,22,28 . Compare the results discussed in A. Nitzan, Chemical dynamics in condensed phases relaxation, transfer and reactions in condensed molecular systems: page 600 Figure 16.9. Eventually, the experimental charge transfer rate constant for the Si-Me-Fc electrode, K ET = 0.98 s −1 , is found to be smaller than the theoretical value of 77.8 s −1 , the latter was obtained by application of the Marcus theory, which represents a theoretical upper ideal limit.
As a whole, it must be noted that surface engineering with ferrocene immobilized moieties was successfully achieved on H-terminated silicon substrate, a route which is promising in building and tailoring nanoscale devices for applications in the field of molecular electronics, spintronics, photonics, catalytic properties.

Materials and Methods
Materials and Reagents. All the reagents used in this work are from Sigma, unless otherwise stated (compare the synthetic details in the Supporting Information), and used as received without any further purification. Highly n-doped silicon substrates (degenerate) with a resistivity <0.010 ohm.cm and thickness in the 475 to 525 µm range, were purchased from Siltronix.

Preparation and functionalization of H-terminated Si(111) surface. Silicon substrates were cleaned
by sonication in n-hexane, acetone, then ethanol (15 min each) and then, dried under a N 2 stream. Hydrogen terminated Si(111) surfaces [H-Si(111)] were prepared following the procedure described by Dumas and Kato 53,54 . A series of acid and basic hot bath cleaning steps were performed followed by a final NH 4 F treatment 54 . Functionalization of the H-terminated Si(111) surface was carried out using a standard Schlenk-line procedure for reaction in inert atmosphere. Photo-induced grafting of was carried out by covering the H-terminated Si(111) surface with Fc-CH 2 OH and irradiating the sample for 1 h with a power of ~35 mW cm −2 using a quartz-iodine lamp with the main emission peak at 236 nm. The flask was heated above the melting point of Fc-CH 2 OH, at 56 °C.
Electrochemical measurements. Cyclic voltammetry (CV) measurements were performed using CHI 660 A and Metrohm Autolab Pgstat 128 N potentiostats, in a typical three-electrode electrochemical cell arrangement. Fc-CH 2 -OH modified H-terminated Si(111) substrates used as a working electrode (active area = 78 mm 2 ), a Pt wire and a Saturated Calomel Electrode Hg/Hg 2 Cl 2 /KCl sat (SCE) electrodes used as the counter and reference electrodes, respectively. Ohmic contacts were made by rubbing indium-gallium (In-Ga) eutectic alloy (495425 Aldrich) on the back of the previously scratched to further improve the contact 51 . 0.1 M tetrabutylammonium tetrafluoroborate (TBATFB) in acetonitrile (ACN) was the base electrolyte. X-ray photoelectron spectroscopy (XPS) measurements. For XPS measurements, Mg K α radiation was used from a dual anode non-monochromatic X-ray source (VG-XR3) operated at 15 mA, 15 kV. Spectra were acquired with double-pass cylindrical mirror analyzer (PHI-15-255G) at constant pass energy of 50 eV.