Mechanistic Investigations into the Selective Reduction of Oxygen by a Multicopper Oxidase T3 Site-Inspired Dicopper Complex

Understanding how multicopper oxidases (MCOs) reduce oxygen in the trinuclear copper cluster (TNC) is of great importance for development of catalysts for the oxygen reduction reaction (ORR). Herein, we report a mechanistic investigation into the ORR activity of the dinuclear copper complex [Cu2L(μ-OH)]3+ (L = 2,7-bis[bis(2-pyridylmethyl)aminomethyl]-1,8-naphthyridine). This complex is inspired by the dinuclear T3 site found in the MCO active site and confines the Cu centers in a rigid scaffold. We show that the electrochemical reduction of [Cu2L(μ-OH)]3+ follows a proton-coupled electron transfer pathway and requires a larger overpotential due to the presence of the Cu-OH-Cu motif. In addition, we provide evidence that metal–metal cooperativity takes place during catalysis that is facilitated by the constraints of the rigid ligand framework, by identification of key intermediates along the catalytic cycle of [Cu2L(μ-OH)]3+. Electrochemical studies show that the mechanisms of the ORR and hydrogen peroxide reduction reaction found for [Cu2L(μ-OH)]3+ differ from the ones found for analogous mononuclear copper catalysts. In addition, the metal–metal cooperativity results in an improved selectivity for the four-electron ORR of more than 70% because reaction intermediates can be stabilized better between both copper centers. Overall, the mechanism of the [Cu2L(μ-OH)]3+-catalyzed ORR in this work contributes to the understanding of how the cooperative function of multiple metals in close proximity can affect ORR activity and selectivity.


General
All manipulations were performed under inert atmosphere using standard Schlenk techniques or inside of a N2-filled MBRAUN LABmaster DP glovebox using anhydrous solvents and reagents, unless noted otherwise. Glassware was dried at 130 ˚C prior to use. Solvents were collected from an MBRAUN MB-SPS-800 solvent purification system and stored over 4 Å molecular sieves, except for CH2Cl2, which was stored over 3 Å molecular sieves. Acetonitrile was stored over 3 Å molecular sieves before being passed over a pad of activated alumina. Deuterated solvents were obtained from Cambridge Isotope Laboratories, degassed, and stored over 3 Å molecular sieves when necessary. D2O for the kinetic isotope experiments was obtained from Eurisotop (99.9 % D). All commercial reagents were used as received and were obtained from Sigma Aldrich, Strem or Acros. All electrolyte solutions were prepared using high purity salts, NMR data was recorded on an Agilent MRF 400 equipped with a OneNMR probe and Optima Tune system or a Varian VNMR-S-400 equipped with an AutoX probe. All resonances in 1 H-NMR were referenced to residual proteo solvent peaks (7.26 for CDCl3, 7.16 for C6D6, 5.32 for CD2Cl2 and 4.79 for D2O). The respective 19 F NMR spectra were referenced employing absolute referencing using the 1 H NMR spectrum of the same sample. EPR spectra were recorded on a Bruker EMXPlus X-band spectrometer. UV-Vis data was recorded on a PerkinElmer Lambda950 spectrophotometer or an Agilent Varian Cary 50 spectrophotometer. The pH of the electrolyte solutions was altered by addition of diluted solutions of NaOH·H2O (Merck, Suprapur 99.9%) or H2SO4 (96%, Merck, Suprapur ). The pH of all electrolyte solutions was determined on a HI 4222 pH meter (Hanna Instruments). The concentration of H2O2 in aqueous solution was determined with H2O2 test strips by use of photometric determination with the Merck Reflectoquant system. S4

Electrochemical experiments
All electrochemical measurements were recorded using Autolab PGSTAT 12,204,   LaboPol-30 polishing machine with diamond polish (1.0 µm, 3 minutes) and silica suspension (0.04 µm, 3 minutes) on polishing cloths (Dur-type). Afterwards, the electrode was sonicated in Mili-Q water for 15 minutes. Before every R(R)DE experiment, the electrolyte solution was bubbled with either Ar or O2 gas for at least 25 minutes. During the experiment the gas was bubbled through the solution.

Synthesis of 2,7-bis(chloromethyl)-1,8-naphthyridine
The compound was synthesized according to a modified literature procedure. 1 A suspension of trichloroisocyanuric acid (1.03 g, 4.42 mmol) in chloroform (15 mL) was added in portions to a cooled (0 °C), stirring solution of 2,7-dimethyl-1,8-naphthyridine (1.00 g, 6.32 mmol) in chloroform (15 mL), resulting in a turbid, canary yellow mixture. After stirring for 15 min, the cooling was removed, and the mixture was allowed to warm up to room temperature. After 50 min the starting material was consumed (monitored by TLC) and water was added (20 mL) causing a slurry to form. The solids were removed using a glass frit and the filtrate was washed with water (2 x 20 mL) and brine (20 mL). The organic layers were dried over Na2SO4, filtrated and concentrated under vacuum to a pale-yellow solid. The compound could be purified by flash column chromatography (silica, DCM:MeOH, 97:3), which after drying over Na2SO4, filtration and drying under vacuum resulted in the formation of the title compound as an off-white powder in 33% yield (0.48 g, 2.09 mmol). Spectroscopic data is consistent with the literature. 1 Note: The monochlorinated compound is also formed in appreciable yields during the reaction and can be isolated during flash column chromatography and chlorinated further.

Synthesis of [BPMANCu2](OTf)2
The compound was synthesized according to a modified literature procedure. 4

Deposit tests in CV
The possible formation of deposited species on the GC electrode and their electrocatalytic activity were studied. To do so, the same procedure was followed at all times: 3+ was measured under non-catalytic (Ar atmosphere) or catalytic conditions (O2 atmosphere / 1.1 mM H2O2).
2. The electrode was thoroughly rinsed with Mili-Q water.
3. The rinsed electrode was transferred to a blank solution in the absence of catalyst and a CV was measured under non-catalytic (Ar atmosphere) or catalytic conditions (O2 atmosphere / 1.1 mM H2O2). Table S1 were carried out and the results are discussed in the text hereafter.

Figure S9c
Catalytic conditions (O2 atmosphere), 0.3 mM catalyst, PB pH7 Catalytic conditions (O2 atmosphere), PB pH 7 Deposit formed that is active for the ORR

Figure S9d
Catalytic conditions (1.1 mM H2O2), 0.3 mM catalyst, PB pH 7 Catalytic conditions (1.1 mM H2O2), PB pH 7 Deposit formed that is active for the HPRR Figure S10a Non-catalytic conditions (Ar atmosphere), 0.3 mM catalyst, PB pH7 Non-catalytic conditions (Ar atmosphere), Acetate Buffer pH 4.8 or PB pH7 Redox couple of the adsorbed species shifts in the same manner as the redox couple of [Cu2L(μ-OH)] 3+ when the pH is changed Figure S10b Non-catalytic or catalytic conditions (Ar or O2 atmosphere), 0.3 mM catalyst, PB pH7 Catalytic conditions (O2 atmosphere), PB pH 7 Deposit formed under catalytic conditions is more active for the ORR

RHE)
Catalytic conditions (O2 atmosphere), PB pH 7 The activity of the deposition that forms during linear-sweep voltammetry is low compared to the activity of the deposit that forms during 3 CV scans. On the contrary, the deposit that forms on the electrode during catalysis of the ORR by [Cu2L(μ-OH)] 3+ is clearly more catalytically active than the bare GC electrode (see Figure S10b).
Next, we investigated the ORR activity of the deposit that forms on the electrode surface when [Cu2L(μ-OH)] 3+ is measured in presence of O2. These CV measurements were recorded at low catalyst RHE and the rinsed electrode was transferred to a blank solution in presence of O2 ( Figure S12). From this measurement it is clear that the deposit that forms on the electrode during the linear sweep voltammetry is less active than the homogeneous catalyst in solution and is also less active than the deposit that forms in a similar experiment over 3 CV scans ( See Figure S9c). This observation indicates that the contribution S23 of deposit is minimal in the first part of the catalytic CV measurement of the homogeneous catalyst and is an indication that the deposit builds up over multiple CV scans. S24 below 0.2 V vs RHE. It is therefore assumed that most deposits are formed after the peak in the catalytic current and hence, the catalytic activity of the first scan is not influenced by active deposits that can form.

Scan rate dependence and diffusion coefficient calculations
The [Cu2L(μ-OH)] 3+ redox couple was recorded at varying scan rates between 10 mV/s and 500 mV/s ( Figure S14a)

CV measurements
The CV response of [Cu2L(μ-OH)] 3+ was measured in a buffered Britton-Robinson solution over a wide pH range (See Figure S15a). A corresponding plot of the E1/2 of these CV scans as a function of the pH is shown in Figure S15b. In general, the data points can be reproduced upon changing the pH from low to high and back.As seen in this figure, four points clearly resulted in a species with a very different E1/2 value than expected. For clarity, these points were not included in the differential pulse voltammetry (DPV) analysis discussed below. Figure S16). The reduction peak shifts to more negative potentials upon changing the pH of the solution from low to high. In case of the oxidation, two peaks are observed. At low pH, these two peaks have approximately the same intensity, and their overlap causes the formation of a third peak in between the other two peaks at certain pH values. Upon changing the pH from low to high, the oxidation peak at high potential decreases, while the peak at lower potential increases and shifts to more negative potentials. This observation indicates the presence of an equilibrium between two different oxidation events that are strongly dependent on the pH. The oxidation peak at lower potential can be found at approximately the same potential as the reduction peak, indicating that these peaks belong to the reduction and oxidation of the same species.

Experiment with [Cu2L] 2+ and H2O2
H2O2 (from a 35% aqueous solution, 1.0 μL, 33 μmol) was diluted 1000-fold with D2O and degassed.      When a solution of [Cu2L] 2+ in acetonitrile is exposed to air in the presence of TEAPF6, a color change of the bulk solution from yellow to green is observed, suggesting the formation of a Cu(II) species ( Figure   S29). In the absence of any proton source, the color change is different, and a greyish-green solution forms.

Kobs for ORR and HPRR in different buffers
Values for the kobs were determined from the background corrected currents of the ORR and HPRR measurements under non-substrate limited conditions. The measurements from the concentration dependence studies (Figure 7, main text) were used to determine the values in PB of pH 7. To determine the kobs values in acetate buffer of pH 4.85, measurements with 3 µM catalyst concentrations were recorded. All background-corrected CV measurements in both buffers are shown in Figure S32. The diffusion coefficient of [Cu2L(μ-OH)] 3+ in acetate buffer was determined in the same manner as described in section SI 5, and is 9.2  10 -7 cm 2 /s for the reduced species. The final values of kobs in Table S2 were calculated as the average value determined for all catalytic waves in Figure S32.

HPRR measurements in D2O
Before every electrochemical experiment in D2O, all glassware was dried in an oven at 140 °C overnight.
The GC electrode was polished as described before and sonicated in D2O. The counter electrode was flame annealed and rinsed with D2O before the experiments. The electrolyte solution was prepared from nondeuterated phosphate salts in D2O. The apparent pH of the prepared solutions was determine using a pH meter calibrated with H2O solutions and converted to obtain pH values of 6.99 and 7.02. 5 To prepare the 1.1 mM H2O2 solutions, a small volume of H2O2 was added from a 10 M stock solution to ensure that the proton content was minimal.

Radical trapping experiments
To investigate the catalytic HPRR mechanism of [Cu2L(μ-OH)] 3+ , hydroxyl radical trapping experiments were performed using 5,5-dimethyl-1-pyrroline N-oxide (DMPO). 6 Treating a MeCN solution of [Cu2L] 2+ and DMPO with H2O2 in the absence of oxygen led to a change of color from dark yellow to green. The characteristic resonance of the DMPO-OH radical was not detected in the EPR spectrum, which is indicative of non-Fenton type reactivity taking place (See Figure S34a). As a positive control, an EPR spectrum of a DMPO solution with H2O2 and FeSO4 was recorded, inducing a Fenton reaction and resulting in the characteristic EPR signal of the DMPO-OH radical (See Figure S34b).

RDE CVs and Koutecky-Levich analysis
Catalytic RDE CVs of [Cu2L(μ-OH)] 3+ were recorded at varying rotation rates between 400 RPM and 2800 RPM ( Figure S35a). The corresponding Koutecky-Levich plot ( Figure S35b) shows linearity over the complete range, suggesting that the number of electrons transferred at -0.4 V vs. RHE in the ORR is independent of the rotation rate.

H2O2 selectivity in RRDE
To calculate the ORR product selectivity in RRDE experiments, any contribution from oxidation of the formed [Cu2L] 2+ species to the observed ring current should be measured. By fixing the Pt ring at 0.8 V vs.
RHE, a potential below the oxidation potential of H2O2, but above the oxidation potential of [Cu2L] 2+ , it was observed that oxidation of the catalyst hardly contributes to the ring current ( Figure S36). This denotes that the largest part of the ring current corresponds to the detection of any H2O2 produced in the ORR.

S46
It is important to determine the collection efficiency of the Pt ring prior to every RRDE experiment in order to precisely quantify the H2O2 that is produced in the ORR. The collection efficiency (N) was determined from Equation 5, using the current collected at the ring (iring) and current collected at the disk (idisk) obtained in a CA measurement of the ORR activity of a blank GC electrode at -0.1 V vs. RHE during 5 minutes, as shown in Figure S37. Since GC is a 100% selective catalyst to H2O2, the collection efficiency of the ring can be calculated by dividing the ring current by the disk current. The currents recorded at -0.

S47
The selectivity of [Cu2L(μ-OH)] 3+ to H2O2 during catalysis of the ORR was calculated from Equation 6. To increase the accuracy of the selectivity in the LSV experiment, the catalytic currents were corrected for the average currents measured between 0.8 and 1.0 V vs. RHE. In the same manner, the H2O2 selectivity in CA measurements (Figure S38) was determined by correcting for the average currents recorded at the ring with the GC disk set to 0.8 V vs. RHE during the 60 seconds prior to the experiment. The H2O2 selectivity in Figure 8 (main text) was determined as the average selectivity of the first 60-90 seconds of the measurement the CA measurements in Figure S38 using Equation 6.   3+ solution, after which the RRDE electrode was rinsed with Mili-Q. Thereafter, an RRDE CV of the rinsed electrode was recorded in a blank buffer solution under O2 atmosphere. The rinsed electrode shows a higher ORR activity than the bare GC electrode, but the activity is lower than of the homogeneous catalyst in solution ( Figure S39a). Next the ORR selectivity to H2O2 was determined for the deposit. This shows that the deposit has a similar selectivity as the homogeneous [Cu2L(μ-OH)] 3+ species (Figure S39b).

Stability of H2O2 solutions in the presence of [Cu2L(μ-OH)] 3+
The disproportionation of H2O2 in PB in the presence of [Cu2L(μ-OH)] 3+ was monitored over time and compared to a control solution with H2O2 in the absence of catalyst ( Figure S40a). This indicates that [Cu2L(μ-OH)] 3+ enhances the decomposition of H2O2, but at such low rates that over the time course of an RRDE experiment this will not affect the ORR selectivity. UV-vis spectra of the solution with H2O2 and [Cu2L(μ-OH)] 3+ were recorded simultaneously, showing that [Cu2L(μ-OH)] 3+ itself is stable in the presence of H2O2 (Figure S40b).

General remarks
Calculations were performed using the Gaussian 16 rev. C01 software. 7 The Becke 1988 exchange functional (B3LYP) was used. [8][9][10][11] For the geometry optimizations the redefinition of Ahlrichs split valence basis set (def2SVP) was used on all atoms. 12,13 Starting geometries for the optimizations were obtained from the coordinates of the crystal structures if possible, or by modification of the optimized geometry of the most similar complex. For the single point calculation on the optimized structures, the redefinition of Ahlrichs triple-zeta split valence basis set (def2-TZVP) was used on all atoms. 12 A SMD continuum solvation model for water was used for the best approximation of the experimental conditions. 14 Additionally, Grimme's DFT-D3 scheme for atom-pairwise dispersion correction with the Becke-Johnson damping(GD3BJ) was used for all atoms in every calculation. 15