Cation Effects on the Acidic Oxygen Reduction Reaction at Carbon Surfaces

Hydrogen peroxide (H2O2) is a widely used green oxidant. Until now, research has focused on the development of efficient catalysts for the two-electron oxygen reduction reaction (2e– ORR). However, electrolyte effects on the 2e– ORR have remained little understood. We report a significant effect of alkali metal cations (AMCs) on carbons in acidic environments. The presence of AMCs at a glassy carbon electrode shifts the half wave potential from −0.48 to −0.22 VRHE. This cation-induced enhancement effect exhibits a uniquely sensitive on/off switching behavior depending on the voltammetric protocol. Voltammetric and in situ X-ray photoemission spectroscopic evidence is presented, supporting a controlling role of the potential of zero charge of the catalytic enhancement. Density functional theory calculations associate the enhancement with stabilization of the *OOH key intermediate as a result of locally induced field effects from the AMCs. Finally, we developed a refined reaction mechanism for the H2O2 production in the presence of AMCs.

bottom part of the cell (10 mL/min).The RRDEs were polished with an 0.05 µm aluminium oxide polishing suspension (Buehler) and cleaned in an ultra-sonication bath with isopropanol and ultrapure water.The electrochemical tests were controlled by a Biologic SP 150 in combination with the EC-Lab software.All potentials were reported vs RHE scale and manually IR corrected using equation S. The Ohmic resistance was obtained by galvanostatic electrochemical impedance spectroscopy (GEIS) at 0 mA.The frequency of the sine function used in the study was in the range 1 to 100000 Hz, and the amplitude was set to 100 µA.
To quantify the selectivity of the reduction towards the 2e -ORR, the H2O2 molar fraction selectivity was applied and calculated accordingly to Equation 2  Afterwards the electrolyte was exchanged to a solution of 0.1 M H2SO4 + 0.2 mM H2O2, without exposing the cleaned electrode to atmosphere.In total 5 L of 0.1 M H2SO4 + 0.2 mM H2O2 were circulated during the experiment, using a peristaltic pump.Therefore, a constant concentration of H2O2 can be assumed during the experiment.Additionally, the concentration of H2O2 before and after the experiment was determined by the spectrophotometric analysis with titanium oxysulfate. 1 Only a negligible change in the H2O2 concentration was found.Figure S2 a) shows the ring current density jring after applying a constant potential of 1.2 VRHE over 15.5 h.The decrease in jring with time can be attributed to anion poisoning, which in turn decreases the collector efficiency for the oxidation of H2O2. 2 After additional ring cleaning (30 cycles.200 mV/s, 0.06-0.76VRHE) the jring increased drastically (Figure S2 b)).Nevertheless, the authors chose not to include additional ring cleaning in the main experiments (Figure 1 a-f)), in order to avoid electrodeposition of Pt on the GC.Identical trends were observed in 0.1 M H2SO4 +0.05 M K2SO4 + 0.2 mM H2O2.

Supporting Note 3:
While applying a constant disk potential of -0.4 VRHE the cationic enhancement effect on the 2e -ORR is apparent.In 0.1 M H2SO4 the absolute value of the resulting currents decrease with time, whereas in 0.1 M H2SO4 + 0.05 M K2SO4 the absolute value of the resulting currents in the first minute's decrease strongly and later increase steadily.Interestingly, resulting in similar X in 0.1 M H2SO4 and 0.1 M H2SO4 + 0.05 M K2SO4, implying higher H2O2 production rates in the presents of K + cations but a constant ORR selectivity.Please note, that the value of the X can be misleading due to the accumulation of H2O2 caused by a partial oxidation at the Pt-ring (starting with a collector efficiency of 37 % but decreasing due to Pt poisoning) and the Pt-ring poisoning (Supporting Note 2) itself.To account for the H2O2 accumulation in the cyclovoltammetry experiments the authors performed a background subtraction of the ring current in the region where no disk current was detected.This approach is not applicable for chronoamperometry.

Supporting Note 4:
With and without the presence of K + metal cations, small Fe and Pt particles (100 nm -1 µm) were found.The origin of the iron contamination is likely to be the threads of the RRDE and the inner parts of the RRDE tips which are made of stainless steel.Pt particles could originate from dissolution of the Pt-ring and the consequently electrodeposition of Pt-particles on the GC electrode.Although, it is difficult to distinguish between Pt particles from electrodeposition and Pt abrasion from the polishing procedure of the electrodes, since Pt particles were found before and after electrochemistry.
However, since the experiments were repeated multiple times and the initial activity overlapped each time the influence of Pt abrasion seems to be negligible.Interestingly, although comparable amounts of Fe and Pt particle were found while performing the electrochemistry in K + containing electrolytes, no significant increase in the activity for the side reactions was found.

Supporting Note 5: In situ XPS measurements
In situ XPS measurements were carried out at the ISISS beamline NAP-XPS end station in BESSY II.For all measurements the in situ membrane-electrode flow cell developed by the Electronic Structure group at the Fritz-Haber Institute 3 , was used.The flow cell allowed the combination of in situ liquid phase electrochemical experiments with XPS measurements.The flow cell utilizes a proton exchange membrane (PEM, Nafion N117) in the combination with a transparent 2D working electrode (trilayer graphene TLG).The graphene layer was directly used as the catalyst.Utilizing the liquid confinement effect, the catalyst is in constant contact with the electrolyte.The main body of the cell is made of polyether ether ketone (PEEK).A Pt wire is used as the counter electrode and an Ag/AgCl electrode as the reference electrode.The TLG is used as the catalyst, current collector, transparent electrode for the photoelectrons and as an evaporation barrier for the liquid electrolyte, enabling the formation of a confined thin-film of liquid environment.The electrolytes were pumped through the flow cell using a peristaltic pump.

Working electrode fabrication
The polymer support of the TLG was removed by carefully placing the samples under MilliQ water, resulting in a floating graphene layer and a sacrificial polymer layer on top.The TLG was afterwards transferred onto the PEM.The TLG on the PEM is stabilized by Van der Waals interaction.The sacrificial layer on the TLG was removed by washing in ethyl acetate.The resulting TLG on PEM was rinsed with MilliQ water, dried at room temperature for 24 h and stored under N2.
Survey and high-resolution spectra were acquired at pass energies of 50 eV and 20 eV and an excitation energy of 1100 eV and 1000 eV respectively.Data analysis and quantification were performed using CasaXPS software.A Shirley background subtraction was applied to the C 1s -K 2p spectra.Fitting of the C 1s -K 2p spectra was performed using a model which includes 10 singlets corresponding to different C 1s species and one doublet corresponding to the K + species.The different C species are C sp2 , C sp3 , C-O(H), C=O, O-C=O, and C1s satellites, which account for the graphene and adventitious carbon with different functional groups, and CF2, CFO, CF2O, and CF3, which account for the Nafion membrane.The K /(C + K) molar ratios were calculated based on the integrated K 2p peak area, the integrated area of the whole C 1s-K 2p region and the corresponding cross sections at 1000 eV.E WE @ 0.49 V RHE , 0 h E WE @ 0.49 V RHE , 2.5 h E WE @ 0.49 V RHE , 4.5 h E WE @ -0.61 V RHE , 5 h E WE @ -0.61 V RHE , 6.5 h E WE @ -0.61 V RHE , 8.5 h E WE @ -0.61 V RHE , 10.5 h E WE @ 0.49 V RHE , 11 h E WE @ 0.49 V RHE , 12.5 h E WE @ 0.49 V RHE , 14.5 h

Supporting Note 6: Computational Details
The density functional theory (DFT) calculations were carried out with the Quantum ESPRESSO 4 simulation package, utilizing GGA-PBE functionals to describe the exchange-correlation energy 5 and SSSP pseudopotentials to account for the core electrons 6 .A 20 Å vacuum was set along the surface perpendicular direction to avoid interactions between periodic images and the Brillouin zone was sampled with a Monkhorst-Pack k-point grid of 2x2x1 was employed for the supercell ORR calculations.In this work, the system was modelled by taking graphene sheets with 8x8 rings and introducing defects in the network, as shown in Figure S13.The defects were chosen according to the following criteria: with experimental evidence of being found in carbon-based materials, and previous theoretical calculations indicating their catalytic activity towards the 2e -ORR. 7The computational hydrogen electrode (CHE) framework was employed to describe the ORR thermodynamics. 8In this model, the free energy of an electron-proton pair is equivalent to half of an H2 molecule considering a system in equilibrium at pH = 0 and the ORR intermediate species adsorption energies (*OOH, *OH, and *O) are calculated considering gaseous H2 and H2O molecules as references. 8Since the *OOH intermediate species is key for the H2O2 electrogeneration, ΔG*OOH was used as the catalytic activity descriptor. 91][12] The Gibbs free energy was then calculated for each applied field with respect to the system without any field effects (E = 0 V/Å), treating vibrational and entropic contributions as constant through the selected potential range. 10,11 he limiting potential (UL) is defined by the lowest potential value in which ΔG is still negative for all reaction steps.The electric field depends on the dipole moment and the polarizability.Therefore, different graphene flakes (with size from 42 to 110 atoms) were investigated to evaluate the effect of their size and thus the K + -coverage/concentration on the dipole moment and the polarizability.In all cases K + (H2O)7 cluster were applied at exactly the same distance from graphene surface.Within one layer of graphene the dipole moment is almost independent from the graphene flake size (Table S3).This can be explained by the fact that already for the smallest tested flake, the valence electron of potassium is completely (based on Mullicken charges and spin densities) transferred to the nanographene, thus the charges and distances (that define dipole) stay constant.However, polarizability is strongly affected by the size of the graphene flakes (Table S3).Coordinates of calculated systems as well as pictures of them given in Table S4-8.Applying the BHHLYP functional resulted in very similar values and are therefore not shown.We would like to underline units of all calculated parameters (Table S3), as a.u.definitions from ORCA output might be confusing.

Figure S1 :Figure S2 :
Figure S1: Determination of the potential of zero charge (PZC) of GC: Exemplary cyclovoltammetry in capacitive region of glassy carbon in 0.1 M H2SO4 at a) 5 mV/s and b) 200 mV/s in Ar-saturated electrolyte at 1600 rpm.At pure capacitive behavior with no faradaic charge PZC is located at Icathodic = Ianodic, dashed red line at I =0 mA.The PZC of GC in 0.1 M H2SO4 and 0.1 M H2SO4 + 0.05 M K2SO4 was determined to be 0.3 and 0.315 VRHE correspondingly.

Figure
Figure S2 a) and b) show the Pt-ring poisoning of the RRDE electrode under a constant potential of 1.2 VRHE over time at 1600 rpm.Prior each experiment the Pt-ring was cleaned while cycling between 0.060 and 1.5 VRHE in 0.1 M H2SO4 (Ar -saturated) until the characteristic Pt-features were reached.

Figure
Figure S4: a) Picture of RRDE after 17h with UTP below PZC in 0.1 M H2SO4 + 0.05 M K2SO4.b) SEM/ EDX images of GC shown in a).c) SEM/ EDX images of GC after 17h with UTP above PZC in 0.1 M H2SO4 + 0.05 M K2SO4.RRDE tip was blow dried in N2 directly after both electrochemical experiments.

Figure S5 :
Figure S5: Effect of the different K + concentrations on 2e -ORR in 0.1 M H2SO4 at GC at 1600 rpm and 5 mV/s.CVs after constant activity was reached for 0.1 M H2SO4 + x K2SO4 with x= 0.1, 0.05 and 0.01 M.

Figure
Figure S7: a) SEM/ EDX mapping of Fe particles b) SEM image/ EDX spectra of Pt particle on GC after electrochemistry with UTP below PZC in 0.1 M H2SO4.RRDE tip was blow dried in N2 directly after electrochemical experiment.

Figure S9 :
Figure S9: XPS survey spectra of trilayer graphene on Nafion N117 with identification of the observed core levels.Survey spectra were acquired at pass energy of 50 eV and an excitation energy of 1100 eV.

Figure S10 :Figure S11 :
Figure S10: XPS spectra of trilayer graphene on Nafion N117 (Spot No.2) of K 2p and C 1s region.Highresolution spectra were acquired at pass energies of 20 eV and an excitation energy of 1000 eV.

Figure S12 :
Figure S12: Averaged K/(C + K) molar ratios of whole sample over time as function of applied potential for short electrochemical protocol.

Figure S13 :
Figure S13: Typical defects and functional groups in carbon-based materials used for the computational calculations.

Table S1 :
Coordinates of the optimized surface models.