Mechanically Interlocked Carbon Nanotubes as a Stable Electrocatalytic Platform for Oxygen Reduction

Mechanically interlocking redox-active anthraquinone onto single-walled carbon nanotubes (AQ-MINT) gives a new and advanced example of a noncovalent architecture for an electrochemical platform. Electrochemical studies of AQ-MINT as an electrode reveal enhanced electrochemical stability in both aqueous and organic solvents compared to physisorbed AQ-based electrodes. While maintaining the electrochemical properties of the parent anthraquinone molecules, we observe a stable oxygen reduction reaction to hydrogen peroxide (H2O2). Using such AQ-MINT electrodes, 7 and 2 μmol of H2O2 are produced over 8 h under basic and neutral conditions, while the control system of SWCNTs produces 2.2 and 0.5 μmol, respectively. These results reveal the potential of this rotaxane-type immobilization approach for heterogenized electrocatalysis.


FigureS 1: Synthesis reaction scheme of mac-AQ.
Anthraflavic Acid (0.5 g, 2.08 mmol, 1 eq) was dispersed with sonication in dry dimethylformamide (DMF) (21 mL, 0.1 M). Then, dry carbonate potassium (K2CO3) (0.86 g, 6.24 mmol, 3 eq), 8-Bromo-1-octene (0.7 mL, 4.16 mmol, 2 eq) and catalytic amount of sodium iodide (NaI) were added and the mixture was refluxed overnight under N2 atmosphere. The next day, the crude of the reaction was allowed to rt and was poured into ice-cold 1 M HCl and the precipitate was filtered. The solid was redissolved in dicloromethane (DCM) and washed with water twice times. The organic phase was dried over Na2SO4 and the solvent was evaporated under reduced pressure. Finally, the crude was purified by flash chromatography (DCM) to obtain the product in 86% yield. 1   S-4

Synthesis and characterization of MINT
The (6,5)-enriched SWCNTs purchased from Sigma Aldrich were purified previously. 100 mg of SWCNTs were suspended in 70 mL of 35 % HCl, and sonicated for 10 min. The mixture was poured in 200 mL of MQ water and filtered through a polycarbonate membrane of 0.2 µm pore size. The solid was washed with water to neutral pH and then dried in an oven at 350 ⁰C for 30 min. Pristine plasma-purified SWCNTs were used without previous purification.
The pristine (10 mg) were suspended in 10 mL of tetrachloroethane through sonication (10 min) and

Determination of hydrogen hydroxide (H2O2) production
The detection of H2O2 was done accordingly to recent literature reports 1,2 . Mixtures in a 1:1 ratio of 4 mM p-nitrophenyl boronic acid (pNBA, Sigma Aldrich, >95%) in DMSO (VWR) and a 150 mM Na2CO3/NaHCO3 (Fluka, >99.5%; Sigma Aldrich, 99.7-100.3%)) buffer solution pH 9 were mixed with the sample. After 36 min incubation at room temperature under dark condition, the absorbance at 411 nm was recorded using a Thermo Fischer Multiskan Go Microplate Spectrophotometer. The amount of H2O2 was determined using a calibration curve made from H2O2 standard solution (Merck, 30%) ( Figure S7). The measured absorbance was subtracted with the absorbance of blank sample (incubation of deionized water with chromophore). The subtracted value, Δabsorbance, was used for the quantification. S-8

Homogeneous CV investigations
As mentioned in the article, an octyloxy-AQ derivative was chosen to compare it to the AQ-MINT electrochemical experiments. In the following Figure S11 the CV of 2.5 mM AQ is compared with the CV of 1mM octyloxy-AQ in 0.1 M TBAPF6 in MeCN solution. The CV curves shown in Figure S11 show, that the alkoxy substituents shift the first reduction wave cathodically by 130 mV whereas the second reduction wave is nearly unaffected.

CV study of SW-CNT
Complementary to Figure 2a  As can be seen from FigureS 12, also pristine SWCNT show a reductive step upon O2 addition, which is already a hint for oxygen reduction S-9

Cycle stabilitynon-aqueous conditions
In the following Figure S13, the cycle stability of AQ-MINT in the organic conditions of 0.1M TBAPF6 in MeCN is shown. In general, a similar behaviour like in aqueous solution (see Figure 3) was observed. The lower jP can be explained by the fact that, in organic solvents in this potential range only a 1-electron reduction process is taking place.

Kinetic investigations
Regarding the cycle stability test, AQ-MINT was reasonably stable, while AQ@SWCNT suffered from severe loss of jP over 50 cycles. Nevertheless, kinetic studies of both AQ-MINT and AQ@SWCNT were performed by CV with varied scan rate (Figure 3a and b). Due to the fact that especially the AQ@SWCNT sample with the 15% loading (and also the AQ-MINT sample with 13%) suffered from significant loss of jP during the kinetic studies, from continuous delamination over the experiment time, no further kinetic data could be extracted thereof. The investigation of samples with lower loadings proved sufficient stability, therefore the mentioned kinetic studies and analysis could be determined.
In the following Figure  The curves shown in Figure S14 were analysed concerning peak currents jP's and the resulting plots as are shown as insets. In both cases, a linear correlation of the jP with the scan rate (v) was observed.
Also, the integrated peak charges were calculated where in case of both AQ-MINT batches quite stable values of, in case of both AQ@SWCNT due to degradation over the cycles only decreasing values were observed.

Further investigations on O2 reduction with AQ-MINT and SWCNT
In order to prove, that the AQ-MINT is stable on the electrode over the time of the electrolysis, in the following Figure S15, CV curves before and after were compared: From the CV in Figure S15 it can be stated, that the AQ-MINT is very stable under the conditions of electrolysis and shows just a slight decrease in jP of 6.3%.
All details on the kinetic parameters like the total moles of H2O2 produced, the rate and the Faradaic efficiency (FE) of oxygen reduction electrolysis at different conditions are displayed in the following Table S1: Although the lack of the exact electrode area hinders further analysis of the slope in Koutecki-Levich-Plot, still the values of iK correlate with the rate constant k, as shown in eq. 2: As those values of iK refer to the exchange current without any convection effects, it can still be regarded as value for comparing the three systems as they arrange on the electrode surface.
S-13 differ but come to quite close values at more negative potentials. In all potentials regarded, AQ-MINT shows the highest iK which can be regarded as a hint for a higher electrocatalytic activity.
As a result, we propose that the different CNT samples with and without AQ modification arrange in a different way on the GC surface, which also affects the real, electroactive surface area.
Unfortunately, because of insufficient resolution of the SEM facilities available, we were not able to proof this assumption.