Host-guest interactions on electrode surfaces for immobilization of molecular catalysts

The strategy of anchoring molecular catalysts on electrode surfaces combines the high selectivity and activity of molecular systems with the practicality of heterogeneous systems. The stability of molecular catalysts is, however, far less than that of traditional heterogeneous electrocatalysts, and therefore a method to easily replace anchored molecular catalysts that have degraded could make such electrosynthetic systems more attractive. Here, we apply a non-covalent “click” chemistry approach to reversibly bind molecular electrocatalysts to electrode surfaces via host-guest complexation with surface-anchored cyclodextrins. The host-guest interaction is remarkably strong and allows the flow of electrons between the electrode and the guest catalyst. Electrosynthesis in both organic and aqueous media was demonstrated on metal oxide electrodes, with stability on the order of hours. The catalytic surfaces can be recycled by controlled release of the guest from the host cavities and readsorption of fresh guest. This strategy represents a new approach to practical molecular-based catalytic systems.


Introduction
Molecular electrocatalysts can exhibit surprisingly high activities and selectivities that are unmatched by most heterogeneous catalysts. 1,2 Therefore, the development of robust immobilization strategies for these molecular species on electrode surfaces is of great interest if these molecular catalytic activities and selectivities are to be transferred to more practical heterogeneous electrosynthetic systems, 3 which have already shown promising results for CO2 reduction, water reduction and water oxidation in the context of storage of renewable energy. [4][5][6] Over the past decades, many immobilization strategies have been developed, 7 which can be categorized into covalent binding (achieved by adding anchoring groups such as carboxylate to the catalysts), 8 non-covalent binding (pi-stacking on carbon-based electrodes) 9,10 and polymerization-based binding. 11 Here, we report a new strategy for surface immobilization of molecular electrocatalysts, which relies on a noncovalent "click" chemistry approach to bind molecular species in well-defined sites on electrode surfaces. 12 The binding of electroactive molecular guests into molecular pockets by means of host-guest complex (HGC) formation has previously been studied by several groups, including those of Stoddart and Kaifer, 13 Reinhoudt 14,15 and Huskens. 16 Liu et al. reported the HGC-formation on gold surfaces with the C60 monoanion as guest, which was found to be electrochemically stable over prolonged durations. 17 Light-induced electron transfer to and from dye molecules bound via the HGC approach was also demonstrated by Freitag and Galoppini. 18,19 The group of Sun demonstrated the use of HGC to improve electron transfer between a molecular catalyst and a dye molecule bound onto TiO2. 20 Among the diverse class of host molecules, cyclodextrins, cucurbiturils and calixarenes are the most studied for HGC formation on different surfaces. 21,22 For cyclodextrins, the electron transfer rates between the electrode and guest molecules situated inside the surface-bound hosts have been reported to be on the order of 100 s -1 , indicating the potential applicability of these systems for molecular electrocatalysis. 23 It was found that the binding of the guest was not affected by redox processes if the binding units do not undergo oxidation themselves.
Although rapid electron transfer to HGC-bound molecules has been demonstrated, to the best of our knowledge there have been no reports on using this approach to bind molecular catalysts to electrode surfaces, likely because the dynamic HGC formation in solution phase suggests that rapid desorption would occur. 24 We demonstrate, however, that molecular guests bind with surprisingly high stability and show that these are catalytically active. We also show that the binding of the guests is reversible under controlled conditions and exchange of these guests can be induced. We further demonstrate regeneration of the catalytic activity of the electrodes after re-adsorption of fresh electrocatalytic guests, implying the high stability of the host under operational conditions of the bound catalysts.

Results and Discussion
We designed the immobilization system to be as compact as possible, using short tethers to attach the hosts to the electrode surface, ensuring close contact between the electroactive guest and the electrode surface. An aromatic binding unit was chosen as guest binding unit to facilitate electron transfer to the catalytically active site of the guest. With the compact design, displacement of the guests by water or solvent from the opposite face of the host should be inhibited. Higher binding constants of guests in surface-bound hosts compared to the binding in solution have been reported, 14,25 and the increased binding strength was proposed to be related to further interactions between guest and surface. For hosts with long, flexible chains connecting them to the surface, studies with multi-valent guests showed that HGCs could be stable on a multi-hour time-scale. 26

Analysis of host-guest complex formation on electrode surfaces
Following our design criteria, we chose to study the per-thiolated derivate of β-cyclodextrin, 1, on gold as a model for host binding close to the surface (Figure 1 a), as this molecule has been thoroughly studied for its surface-functionalization and surface HGC formation behaviour on gold. 13,17,27,28 As guests, we chose to investigate ruthenium-based electrocatalyst 2, inspired by the class of molecular catalysts that has very recently been shown to be active for ammonia oxidation in organic electrolytes (ammonia being a promising candidate as storage medium for hydrogen). [29][30][31][32] We also investigated structurally similar ruthenium-based electrocatalyst 3, a derivate of which we have reported to be active for water oxidation. 33 As a control, fluorescent platinum complex 4 was synthesized, which served as a non-electroactive guest. The complexes used in this study and their crystal structures are given in Figure 1 b-d. The common design feature in the guest metal complexes is the naphthyl substitution on the terpyridine ligand, which was employed as the binding unit of the guests inside the cyclodextrin cavity. To confirm that binding between the guests and cyclodextrin takes place in solution, fluorescence titrations and NMR experiments were performed (see Figures SI-1 and SI-2 in Supporting Info). A 1:1 binding stoichiometry binding constant for fluorescent guest 4 with β-CD was determined and a binding constant of K11 = 6.5·10 3 M -1 was obtained, which is 2-4 times higher than for other naphthyl-based guests with β-CD. 34,35 The attachment of 1 on gold substrate was investigated by X-ray photoelectron spectroscopy (XPS) ( Figure SI-3) and tip-enhanced Raman spectroscopy (TERS) (vide infra). The spectra indicated the expected attachment of the cyclodextrin host via the thiol groups, as was evidenced by the loss of the ν(S-H) Raman band upon adsorption. The surface coverage of 1 on Au(111) was 35 % as determined by XPS, which is in excellent agreement with previous studies. 13 We note that the maximum coverage of the surface was achieved with dipping times of 10 minutes, with no further increase in 1 surface coverage at longer exposure of the gold substrate ( Figure SI-4).
The formation of HGCs with the gold-bound 1 was investigated using TERS, and DFT calculations were performed to gain a better understanding of this system. 36   Thus, conditions need to be developed to actively remove the guests from the surface-bound hosts for recycling purposes. The desorption of guests under competitive HGC formation in solution has been reported, 26,38 and this strategy was also employed to show that guest binding to the 1 modified gold substrates could be controlled under specific conditions.
Using TERS (Figure 3a), we were able to track the presence of guest 3 on gold functionalized with host 1.
Formation of the HGC on the Au surfaces was visualized by the appearance of an intense Raman fingerprint of the guest between 1000 and 1600 cm -1 compared to the moderately strong ν(C-H) band of 1 at 2900 cm -1 (cyclodextrins are very weak Raman scatterers). Desorption of the majority of guest was achieved by soaking and sonication of the sample in a concentrated β-CD solution (indicated by the strong decrease in the guest/host Raman band intensity ratio). Re-adsorption of the guest was observed after re-soaking the desorbed sample in a solution of 3. In a separate experiment, we also showed that guest 4 (Pt complex) can be exchanged for a second guest 2 by initial desorption of 4 using concentrated β-CD solution and subsequent incubation in a solution of guest 2 (Figure 3 b). The exchange was enabled by the release of guest 4 from the surface-bound host cavities during the desorption step and cannot be explained with stronger binding of guest 2 compared to guest 4, as the reverse exchange (2 for 4) is also feasible ( Figure SI-10). Interestingly, complete loss of guest signal after the desorption treatment was never observed, which again hints towards the unexpectedly strong interaction between the guest and the 1 functionalized gold. Direct observation of 1 and the HGCs on Au (111) was attempted by STM, however a clear image could not be obtained despite significant efforts ( Figure SI-11).
Studying the electrochemical properties of the HGCs on gold alone was not sufficient to obtain a clear picture of our catalytic system. Gold electrodes are known to undergo surface oxidation, 39 which leads to the desorption of thiol-based adsorbates. 40,41 Furthermore, the relatively low surface area and high activities of gold for many catalytic processes make the analysis of catalytically active adsorbates very difficult. 42 To address these issues and to compliment the array of available analytical techniques to study the surfacebound HGCs, metal oxide (MO) substrates were used, specifically indium tin oxide (ITO), zirconium dioxide (ZrO2), and titanium dioxide (TiO2). As these materials have similar properties in terms of surface-adsorption chemistry, 43 the choice of substrate can be tailored to the application of analytical technique (ITO for electrochemical measurements, ZrO2 for fluorescence and IR, and TiO2 for solid state NMR). Furthermore, by preparing mesoporous layers of the MOs, higher surface areas are easily accessible, allowing for the use of further spectroscopic methods and simplifying the analysis of electrochemical data.

b) Desorption and subsequent exchange of 4 by 2 on host-functionalized gold is confirmed by the shift of the ring breathing modes (RBM) of the metal-coordinated pyridyl rings from 1050 cm -1 for Pt toward 1030 cm -1 for Ru (highlighted in blue). The incomplete desorption of 4 is reflected in the residual intensity of 4 in spectrum II and in the mixed signals of 4 and 2 in spectrum III.
To transfer the HGC chemistry from the gold to the MO substrates, two strategies were employed: direct introduction of multiple suitable binding groups, geminal bisphosphonates (BP), 44 to the host scaffold (host 5) or pre-functionalization of the MO substrates with propiolic acid and subsequent thiol-yne click chemistry on the surface. 45 Both strategies are schematically shown in Figure SI Figure SI-21)). Formation of the host-guest complexes could also be seen during cyclic voltammetry experiments: whereas physisorbed guests rapidly desorb after a few cycles, redox-features of the HGCs could be observed even after repeated cycling ( Figure SI-22).
Exposing mesoporous ZrO2 substrates with HGC-bound guests to the desorption conditions used for the gold (with heating of the solution instead of sonication, to preserve the mesoporous substrate), the removal of guests from the surface could be tracked by fluorescence spectroscopy. The same reversible desorption and re-adsorption of different guests (as was shown for the gold electrode using TERS) indicates that the surface can be modified without degrading the surface-bound hosts ( Figure SI-23).

Catalysis using host-guest complex bound molecular electrocatalysts
Catalyst 2 -designed as a molecular ammonia oxidation catalyst and equipped with a naphthyl binding groupshows the expected homogeneous phase catalytic activity for NH3 oxidation in THF solution around its first oxidation potential (Figure 4 a)

II)/Ru(III) redox peak is observed (green). Upon addition of ammonia, a catalytic onset for ammonia oxidation is observed at the Ru(II)/Ru(III) oxidation potential (magenta). b) CV scans of HGC-bound guests 2 (magenta) and 4 (black) with host 5 on mesoporous ITO show the same catalytic behavior in the presence of the ruthenium catalyst, with onset at the Ru(II)/Ru(III) oxidation potential measured for HGC-bound 2 without NH3 (green). The conditions are the same as for (a). c) Chronoamperometry at 0.08 V vs Fc/Fc + using host 5-bound catalyst guests 2 and 4 (the latter as a non-catalytically active reference). Readsorption of guest 2 after 30 minutes of catalysis into the host-functionalized electrodes leads to complete regeneration of catalytic activity.
The stability of the HGC-bound electrocatalyst 2 under operation was analyzed by chronoamperometry, revealing a slow decrease in current to roughly 70 % of the initial value over 30 minutes (Figure 4 c), with significantly higher currents observed than for physisorbed 2 without host present ( Figure SI-25). Re-adsorption of fresh 2 regenerated the catalytic activity of the electrode, which clearly indicates that the host molecules on the electrode surface are intact and can be used to bind fresh catalyst guests. The decrease of current could either be due to slow desorption of the guests or degradation of the HGC-bound catalyst. Nevertheless, these results show that our immobilization strategy of molecular electrocatalysts using HGCs can be employed to bind active catalysts to electrodes. Moreover, this strategy was effectively used in organic electrolyte, which is notable considering the general competition for occupying the host cavity between guests and solvent molecules. 46,47 This observation underlines the strong adsorption interaction between the guest molecules and the surface bound hosts.
Electrochemical water oxidation was also attempted using guest 3, which was shown to be active (though sluggish) for this reaction in solution ( Figure SI-26 a). Unfortunately, we could not observe an unambiguous catalytic response for water oxidation versus the background. To nevertheless confirm the activity of HGCbound guests, we were able to demonstrate catalytic oxidation of a water-soluble phosphine to the corresponding phosphine oxide with HGC-bound 3 on ITO as a model reaction ( Figure SI-27). This system also showed similar stability as the case for NH3 oxidation in organic media. Regeneration of catalytic activity upon readsorption of 3 indicates the robustness of the host on the surface ( Figure SI-28).
To confirm that the host is stable during electrocatalysis, we analyzed host 5-functionalized ITO substrates before and after electrochemical measurements (with HGC-bound guest 2). We observed no peak shifts for the carbon, phosphorous or sulfur signals pertaining to host 5 between the two measurements ( Figure 5 and Figure

Conclusion
Host-guest complex formation on electrode surfaces can be used to immobilize molecular electrocatalysts. A detailed study on the formation of such HGCs on gold, along with DFT calculations, shows a strong binding of the guests to the surface-bound cyclodextrin hosts. By demonstrating the catalytic activity of HGC-bound guests in both organic and aqueous electrolyte, our experiments show the versatility of using this approach to immobilize different electrocatalytic guests for operation in different chemical environments. The high stability of the host structures on the electrode surface allows for regeneration of the electrodes by re-adsorption of fresh catalyst guests, which overcomes one of the largest challenges in immobilized molecular electrocatalysis: the electrodes can be reused even when the catalyst molecules have degraded.
Our work sets the basis for the use of HGC chemistry for catalyst immobilization. Future research efforts focusing on the preparation of new host structures and families, expanding the scope of the catalytic reactions and understanding and improving the control over the interactions between guest, host and surface will allow for this immobilization strategy to be used for many different applications.

Methods
Host-functionalization of gold substrates: Gold substrates were used either as prepared (directly after deposition of the gold layer) or cleaned in an O2 plasma before functionalization. Compound 1 was attached to the gold by soaking the substrates in a solution of 1 (0.1 mM) prepared with either DMSO or DMF. Although adsorption was found to take place in less than 5 minutes, the substrates were soaked for 1 h. The substrates were then soaked in pure solvent (5 minutes), MeOH (5 min) and dried under a stream of N2. The hostfunctionalized gold was used immediately or dipped in guest solutions very soon after preparation.