A Three-dimensional Floating Air Cathode with Dual Oxygen Supplies for Energy-efficient Production of Hydrogen Peroxide

The in situ and cleaner electrochemical production of hydrogen peroxide (H2O2) through two-electron oxygen reduction reaction has drawn increasing attentions in environmental applications as an alterantive to traditional anthraquinone process. Air cathodes avoid the need of aeration, but face the challenges of declined performance during scale-up due to non-uniform water infiltration or even water leakage, which is resulted from changing water pressures and immature cathode fabrication at a large scale. To address these challenges, a three-dimensional (3-D) floating air cathode (FAC) was built around the commercial sponge, by coating with carbon black/poly(tetrafluoroethylene) using a simple dipping-drying method. The FAC floated on the water-air interface without extensive water-proof measures, and could utilize oxygen both from passive diffusion and anodic oxygen evolution to produce H2O2. The FAC with six times of dipping treatment produced a maximum H2O2 concentration of 177.9 ± 26.1 mg L−1 at 90 min, with low energy consumption of 7.1 ± 0.003 Wh g−1 and stable performance during 10 cycles of operation. Our results showed that this 3-D FAC is a promising approach for in situ H2O2 production for both environmental remediation and industrial applications.

depth and immature GDL fabrication at a large scale 25,26 . This uneven water infiltration resulted in either water leakage, or a non-uniform current distribution within the cathode, leading to degraded cathode performance 25,26 .
The electrochemical ORR systems often rely on single external O 2 supply methods such as passive oxygen diffusion or active aeration, but neglect the produced O 2 from the anodic oxygen evolution reaction (OER) 10,19,27 . In the two-electron oxygen reduction process, the produced O 2 from the OER anode is half of that consumed at the cathode 10,28 , but this part of O 2 is often wasted. Recently, with the utilization of anodically produced oxygen, we developed an oxygen-self-supplied electro-fenton system with dual cathodes that did not need the feed of external O 2 supply 10 . Similar approaches that utilize the anodically produced oxygen extend the application field of in situ electrochemical H 2 O 2 production 3,29 . In spite of much work on single oxygen supply 18,30 , the electrochemical system to produce H 2 O 2 with multi-oxygen supplies is rarely reported. In addition, it is really significant to figure out contributions of different O 2 supplies to H 2 O 2 production for the future practical applications.
In this study, in order to address the uneven water infiltration issues and enhance the O 2 mass transfer to air cathodes for scale up applications, we developed a floating air cathode (FAC) using the commercially available poly(urethane) (PU) sponge that floats at the solution/air interface for effective H 2 O 2 production without the needs of extensive water-proof measures. The sponge was dipped in the catalyst ink made of carbon black to make it electrically conductive and electrochemically active. The FAC with various dipping times (DTs) was characterized in terms of morphology, mass loading, ohmic resistance (R ohm ), and electrochemically active surface area (ECSA). The FAC performance in terms of H 2 O 2 concentration and normalized energy consumption were also evaluated to find the most cost-effective catalyst loading for the production of H 2 O 2 . We examined the H 2 O 2 production with different relative positions of the cathode in the electrolyte, to characterize the influence of O 2 supply sources (passive diffusion from air, and O 2 produced from the OER anode) to the ORR system.

Methods
Electrode Preparation. The floating air cathode (FAC) was fabricated through a simple and scalable dipping and drying process ( Fig. 1) 31 , around the commercially available PU sponge of 50 pores per inch (ppi) as the support (Hangmei sponge Co., Ltd). Carbon black (CB, acetylene, 50% compressed, Alfra Aesar Co., Ltd) was used both as the catalyst and the conductive layer, with PTFE (60 wt.% dispersion in H 2 O, Sigma-aldrich Co., Ltd) as the binder. CB (300 mg) and 60% PTFE (2 mL) with PTFE/CB ratio of 6:1 were dispersed in 30 mL of ethanol with ultrasonication of 30 min to form a uniform suspension. The sponge was dipped into the suspension and dried in the electric oven at 80 °C for several times 31 . The FAC had a diameter of 4.0 centimeters (cm, projected surface area of 12.56 cm 2 , 1.0 cm in height), and catalyst loading depended on the dipping times. Mixed metal oxides (MMO) mesh (4.0 cm in diameter × 0.1 cm height) was used as the anode.
Operation. All the electrochemical measurements were performed in Petri dishes (6.0 cm inner diameter × 3.0 cm height or 9.0 cm inner diameter × 2.0 cm height, Fig. S1) with 50 mL electrolyte (100 mM Na 2 SO 4 solution). The FAC floated at the solution/air interface, and the MMO anode was placed at the bottom of the reactor and facing the FAC. The electrode spacing between the bottom of FAC and the top of MMO was ~2.0 cm. The electrochemical double-layer capacitance (EDLC) measurements of FACs with different dipping times were carried out using cyclic voltammetry between −0.04 and 0.04 V versus Ag/AgCl with the scan rate of 0.5 mV s −1 , which served as an estimate of the electrochemically active surface area (ECSA) of the solid-liquid interface 32,33 . FACs with different dipping times were measured by linear sweep voltammetry (LSV) from 0 to −1.0 V versus Ag/AgCl with the scan rate of 5 mV s −1 to evaluate their ORR performance. The production of H 2 O 2 was measured at a fixed voltage of 2 V using the potentiostat (VMP3, BioLogic, France). Cathode potentials were measured by the Keithley data acquisition system (2700 multimeter, Keithley, America) versus the Ag/AgCl reference electrodes (all cathode potentials were reported versus Ag/AgCl). Under the same mode with the fixed voltage of 2 V, the FAC of DT6 was chosen to evaluate the stability of H 2 O 2 generation by changing the electrolyte every 90 min for 10 times. The produced H 2 O 2 concentrations were measured with different applied voltages using the FAC of DT6.
To investigate the influence of oxygen supply on system performance, four types of working modes were designed (Fig. S1), including: the mode as mentioned above with the floating air cathode facing the OER anode (Mode 1, M1, oxygen from both air and OER), floating air cathode misplaced with OER anode (Mode 2, M2, oxygen only from air), submerged cathode facing the OER anode (Mode 3, M3, oxygen from OER) and submerged cathode misplaced with OER anode (Mode 4, M4, no continuous oxygen supply). The produced H 2 O 2 concentrations under different working modes were measured, under the same set voltage of 2 V.

Results and Discussion
Characterization of the Floating Air Cathode. The three-dimensional PU sponge that has been commonly used for household cleaning, packaging, filtrating and many other applications, was utilized for the fabrication of the air cathode. The sponge was dipped into a CB/PTFE ethanol solution, allowing the solution to fill the voids and coat the skeletons. The as-formed electrode could freely float on the air/water interface while pumping water in or out the Petri dish, successfully maintaining the O 2 diffusion path from air. Densities of the PU sponge before and after the dipping-drying treatment (0.02 g cm −3 for sponge, 0.05-0.12 g cm −3 for FAC, Table S1) were far less than that of water, guaranteeing the floating property of sponge-based electrodes. The small size of CB powder and strong adhesion of PTFE binder enabled the formation of a carbon "skin" that coated on the sponge surface, as compared with SEM images of the sponge before and after the dipping-drying treatment ( Fig. 2 and S2). With the increase of dipping times, the CB/PTFE layer became thicker and gradually filled macropores of the sponge (Figs 2 and S2). This CB/PTFE coating changed the color of sponge from yellow to black (Fig. 1) and electrical conductivity of the entire matrix from insulative to conductive (Fig. S3). In the FAC, the mass loading of catalyst and ohmic resistance greatly relied on dipping times. With the increase of DTs from one to four, the mass loading of catalyst hugely increased from 25.2 to 92.1 mg cm −2 while ohmic resistance rapidly decreased from 310 Ω to 36 Ω (Fig. 3a). After four times of dipping into the CB/PTFE solution, the mass loading of catalyst and ohmic resistance of FACs tended to be stable. With six times of dipping-drying treatment, the ohmic resistance of PU sponge coated with CB/PTFE mixture was ~31 Ω (Figs 3a and S3b), enabling its function as an electrode in small scale applications.
Effect of dipping times on FAC electrochemical performance. To estimate the electrochemically active surface area (ECSA) of FACs with different dipping times, the electrochemical double-layer capacitance (EDLC) were measured by cyclic voltammetry 35,36 . It showed that the capacitance greatly relied on dipping times of the electrode. With the increasing DTs from one to six, the EDLC of FACs gradually increased from 12.9 to 58.0 mF (Figs 3b and S4), suggesting that the ECSA improved due to the increased catalyst mass loading. Although the measured resistance did not appreciablely decrease after four times, the ECSA still improved with a higher catalyst loading. After six times of dipping treatment, the EDLCs of FACs started to be stable (55.7 mF of DT7 and 57.1 mF of DT8, Figs 3b and S4), attributing to their similar mass loading and ohmic resistance to DT6 (97.9 mg cm −2 and 32 Ω for DT7, 100.5 mg cm −2 and 30 Ω for DT8, Fig. 3a). To evaluate the ORR performance of FACs with various dipping times, linear sweep voltammetry was measured. Due to the same catalytic component, the onset potentials of all FACs were −0.27 V vs. Ag/AgCl (Fig. S5). However, the ORR performance of FACs improved with dipping times. At −1.0 V vs. Ag/AgCl, only −4.6 mA of ORR current was produced on the DT1 FAC, much less than −18.1 mA for DT2. Raising dipping times resulted in increasing ORR current from −31.2 mA for DT3 to −42.5 mA for DT4 and −51.2 mA for DT5. When dipping time was eight, the ORR current reached a maximum of −63.2 mA, slightly larger than those for DT6 (−61.4 mA) and DT7 (−58.3 mA, Fig. S5). The improved ORR performance with DTs was attributed to higher active surface area and lower resistance that resulted from the increased catalyst mass loading. This was consistent with previous studies that mass loading of catalyst, ECSA and electrical conductivity of electrode were positively correlated with the performance of electrochemical systems 34,37 . Thus, when the same set voltage of 2 V was applied to produce H 2 O 2 , cathode potentials became more positive for the FACs with more DTs (Fig. S6). On average, the cathode potential on DT6 FAC was −0.63 V, a little more positive than −0.68 V on DT5 and −0.71 V on DT4. When dipping times were two or three, the average cathode potential was close, approximately −0.75 V (DT3) and −0.78 V (DT2). The FAC that was dipped only once had the most negative cathode potential of −0.88 V (Fig. S6), suggesting the worst cathode performance.
Produced H 2 O 2 were measured using FACs with different dipping times. With the fixed set voltage of 2 V, the cathodic current and H 2 O 2 production showed a positive relationship with the DTs of FACs from once to six times (Figs 3c and S7), attributing to the increasing mass loading, decreasing ohmic resistance and increasing ECSA. When the dipping times were six, the H 2 O 2 concentration reached a maximum of 177.9 ± 26.1 mg L −1 at 90 min, slightly more than 158.7 ± 26.5 mg L −1 for five times (Fig. 3c). Lowering the DTs of FAC led to decreased H 2 O 2 concentration from 122.2 ± 13.1 mg L −1 for four times to 65.6 ± 2.8 mg L −1 for twice (Fig. 3c). Only 17.2 ± 7.2 mg L −1 H 2 O 2 generated if the sponge was dipped into CB/PTFE solution only once (Fig. 3c). The improved H 2 O 2 concentration with DTs from once to six times resulted from larger cathodic current (Fig. S7) and higher coulombic efficiency (Fig. 3d). However, after six times of dipping, the FACs showed similar electrochemical performance, leading to stable H 2 O 2 generation of 177.1 ± 12.2 mg L −1 for seven times and 177.8 ± 12.9 mg L −1 for eight times (Fig. 3c). Normalized energy consumption was evaluated for the H 2 O 2 production using the FACs. The lowest consumed energy of 7.1 ± 0.003 Wh g −1 was obtained at six dipping times for the FACs, close to 7.1 ± 0.01 Wh g −1 at seven times, 7.5 ± 0.35 Wh g −1 at eight times (Fig. 3d). With the decreased dipping times, energy consumption gradually increased from 7.6 ± 0.9 (five times) to 11.9 ± 2.8 (once) Wh g −1 due to the increased ohmic resistance. The FAC was further evaluated by comparing its energy consumption with those of CB-based gas diffusion electrodes (GDEs) reported previously (Table 1) 13,22,27,[38][39][40][41] . The energy consumption obtained here were lower than those with conventional CB-based GDEs (7.45-22.1 Wh g −1 , Table 1) 13,22,27,40 . For the modified cathodes with CoPc or FePc addition, the energy consumption was higher to be 30.8-165 Wh g −1 39,41 . The cathode with tert-butyl-anthraquinone (TBAQ) addition showed superior performance (6.0 Wh g −1 ) 38 due to improved two-electron transfer selection, indicating that the energy consumption in our system could be further lowered with the improvement of catalyst activity.
Effect of applied voltages on H 2 o 2 generation. H 2 O 2 production was measured when applying different voltages using the FAC of DT6. With the set voltage increasing from 2.0 V to 5.0 V, cathodic current gradually increased from 25.6 mA to 170.9 mA (Fig. S8), suggesting that oxygen reduction on the FAC dramatically enhanced. This huge ORR enhancement led to the improving performance of H 2 O 2 generation via two-electron ORR pathway. As a result, the increase of H 2 O 2 concentration with applied voltage was from 177.9 ± 26.1 mg L −1 at 2.0 V to 488.6 ± 77.5 mg L −1 at 3.0 V and 896.4 ± 10.3 mg L −1 at 4.0 V (Fig. 4a). At the set voltage of 5.0 V, the DT6 FAC produced a maximum of 1062.1 ± 79.4 mg L −1 H 2 O 2 within 90 min (Fig. 4a).
Normalized H 2 O 2 generating rate was calculated to evaluate the performance of FAC-based electro-generation system. H 2 O 2 generating rate showed a positive relationship with the applied voltage from 2.0 V to 5.0 V (Fig. 4b).
At the fixed voltage of 2.0 V, H 2 O 2 was generated at the rate of only 0.46-0.60 mg h −1 cm −2 , far less than 1.30-1.76 mg h −1 cm −2 at 3.0 V and 2.38-3.46 mg h −1 cm −2 at 4.0 V (Fig. 4b). When 5.0 V of voltage was applied, the DT6 FAC produced H 2 O 2 at a maximum rate of 2.82-5.80 mg h −1 cm −2 (Fig. 4b). However, within 90 min of electrochemical treatment, H 2 O 2 generating rate gradually decreased from 5.80 ± 0.85 mg h −1 cm −2 at 15 min to 2.82 ± 0.21 mg h −1 cm −2 at 90 min (Fig. 4b) (Table 1) 27,40 due to low solution resistance derived from the extremely small interelectrode gap (2.0-6.5 mm). In addition, concentrated alkaline solution as highly-conductive and high-pH electrolyte was beneficial for electro-reduction of oxygen, resulting in an ultrahigh H 2 O 2 generating  (Table 1) 13 . Therefore, H 2 O 2 generating rate of FAC could be further improved by decreasing solution resistance and increasing electrolyte pH.

Effect of different O 2 supplies and electrode relative positions on the cathode performance.
To investigate the effect of different O 2 supplies on the cathode performance, H 2 O 2 generation under four working modes was measured. These modes represented several different types of oxygen supply approach to the 3-D sponge-based electrode in the ORR system (Figs 5a and S1). The M1 working mode enabled continuous O 2 supply of the FAC from air and anodic OER, resulting in the H 2 O 2 production of 177.9 ± 26.1 mg L −1 within 90 min (Fig. 5b).  Fig. S9). Due to the lack of continuous oxygen supply, the cathodic current gradually decreased from 40 mA to 2 mA in the initial 45 min (Fig. 5c). Compared with submerged cathodes (M3 and M4), the cathodic currents of FACs (M1 and M2) with O 2 supply from air were relatively stable during the whole 90-min operation (Fig. 5c) Fig. 5c). The higher cathodic current with FACs resulted from passive O 2 diffusion from air through the three-phase interface. When cathodes were submerged, the O 2 transfer was greatly hindered in aqueous solution. Therefore, the H 2 O 2 generation from FACs was also much better than that from submerged cathodes. For the FACs (M1 and M2) and submerged cathodes (M3 and M4), the relative position between cathode and anode also affected their electrochemical performance. From a comparison of either M1&M2 or M3&M4, the cathodic current with a facing position was larger than those with a misplaced position (M1 of 20.6 mA > M2 of 11.5 mA and M3 of 5.6 mA > M4 of 1.2 mA, Fig. 5c). The decreased cathodic current with a misplaced position could be attributed to the increased solution resistance, electric field lines change and less oxygen supply without the utilization of OER produced oxygen. When the sponge-based cathodes were misplaced with MMO anodes, the solution resistance increased from 14.7 Ω for M1 to 27.7 Ω for M2 and from 7.4 Ω for M3 to16.2 Ω for M4 (Fig. S10). Therefore, consistent with the trend of cathodic current, the H 2 O 2 production with a facing position was much higher than that with a misplaced position (Fig. 5b).

Stability of the FAC for H 2 o 2 production.
To evaluate the working stability of FAC with dual O 2 supplies, the H 2 O 2 generation with a fixed voltage of 2 V (the actual cathode potential ranged between −0.68 V and −0.56 V, Fig. S6) in a batch of experiments by changing the electrolyte every 90 min was investigated. Under the M1 working mode, the varying range of cathodic current on the FAC was moderate (18-24 mA, Fig. 6a), indicating the relatively stable performance of the FAC. In each run, the linear increase of H 2 O 2 production also reflected the stability and sustainability of two-electron ORR process (Fig. 6b). The yield of H 2 O 2 was almost stable with a fluctuation during ten cycles, and the fluctuating range of H 2 O 2 concentration was from 187 to 237 mg L −1 (Fig. 6b). The fluctuation of H 2 O 2 production could be partially resulted from the variation of ECSA of FAC in different cycles as the cathode floated freely on the solution/air interface. This result showed that the electrochemical ORR system using the FAC was capable of working stably, representing a promising system for energy efficient H 2 O 2 production.

Conclusions
In this study, the floating air cathodes were fabricated using commercially available sponge with a simple dipping-drying method. When the cathode floated on top of the OER anode, it had dual oxygen supply sources both from the air and anodic OER. The optimized FAC produced a maximum H 2 O 2 production of 177.9 ± 26.1 mg L −1 within 90 min, and had a low energy consumption of 7.1 ± 0.003 Wh g −1 and good working stability. The features make the FAC a promising option to in situ H 2 O 2 production in scale up applications for either environmental remediation or industrial applications.