Boosting Nitrate to Ammonia Electroconversion through Hydrogen Gas Evolution over Cu-foam@mesh Catalysts

The hydrogen evolution reaction (HER) is often considered parasitic to numerous cathodic electro-transformations of high technological interest, including but not limited to metal plating (e.g., for semiconductor processing), the CO2 reduction reaction (CO2RR), the dinitrogen → ammonia conversion (N2RR), and the nitrate reduction reaction (NO3–RR). Herein, we introduce a porous Cu foam material electrodeposited onto a mesh support through the dynamic hydrogen bubble template method as an efficient catalyst for electrochemical nitrate → ammonia conversion. To take advantage of the intrinsically high surface area of this spongy foam material, effective mass transport of the nitrate reactants from the bulk electrolyte solution into its three-dimensional porous structure is critical. At high reaction rates, NO3–RR becomes, however, readily mass transport limited because of the slow nitrate diffusion into the three-dimensional porous catalyst. Herein, we demonstrate that the gas-evolving HER can mitigate the depletion of reactants inside the 3D foam catalyst through opening an additional convective nitrate mass transport pathway provided the NO3–RR becomes already mass transport limited prior to the HER onset. This pathway is achieved through the formation and release of hydrogen bubbles facilitating electrolyte replenishment inside the foam during water/nitrate co-electrolysis. This HER-mediated transport effect “boosts” the effective limiting current of nitrate reduction, as evidenced by potentiostatic electrolyses combined with an operando video inspection of the Cu-foam@mesh catalysts under operating NO3–RR conditions. Depending on the solution pH and the nitrate concentration, NO3–RR partial current densities beyond 1 A cm–2 were achieved.

. a)d) Photographs showing consecutive process steps applied for the preparation and application of the Cu-foam@mesh catalyst.    Figure S4. a) Sweep rate dependent cyclic voltammograms (CVs) of the Cu mesh (reference/support) and the electrodeposited Cu foams (5 -60 s deposition time) measured in an electrolyte solution containing 10 mM dimethyl viologen dichloride (DMVCl2; Sigma Aldrich, 98%) as reversible redox probe and 1 mol L -1 Na2SO4 (Sigma Aldrich, ≥99.0) as the supporting electrolyte. Note, the current densities were normalized to the geometric surface area. b) Plots of the (reduction) peak current densities (jpeak) versus the square root of the potential sweep rate (ν 1/2 ). The ECSA values (Table S1)   Ammonia quantification: aliquots of the catholyte were diluted 20 to 100 times with Milli-Q water. Afterward, 2 mL of the diluted catholyte was mixed with 1 mL 0.05 mol L -1 NaClO4 (Sigma-Aldrich, reagent grade), 2 mL of 1 mol L -1 NaOH solution (Sigma-Aldrich, >98.0%) containing 5 wt.% salicylic acid (Sigma Aldrich, ≥99.0%) and 5 wt.% sodium citrate (Sigma Aldrich, ≥99.0%), and 200 L of 1 wt.% sodium nitroferricyanide solution (Sigma Aldrich, ≥99%). After a reaction time of ~1 hour, UVvisible absorption spectra were recorded in the range from 450 to 900 nm with a PerkinElmer Lambda 900 UV-visible/NIR spectrometer. The characteristic absorption maximum of the indophenol derivative was observed at a wavelength of λ = 658 nm (see Figure S5c). The ammonia quantification through the formed indophenol derivative was based on calibration curves (see Figure S5d).    For these control experiments an anodic vertex potential was chosen more negative than the onset of Cu oxidation. The comparison of the CVs recorded in the nitrate and nitrite containing electrolytes confirms that the first cathodic current feature P1 originates from the nitrate→nitrite reduction whereas the main current peak P2 is due to the further reduction of the formed nitrite into ammonia. Note that P1 is absent in the voltammogram when nitrite is the reactant. Highlighted in green are extra cathodic current features often appearing in the reverse potential sweep. These (artifacts) are, however, due to fluctuation of the cathodic currents due to hydrogen bubble formation and release (not reproducible!). These fluctuations occur because of massive gas evolution thus leading to sudden changes of the active surface area and the hydrodynamic conditions. Figure S10. a-e) Optical micrographs of the Cu-foam(30 s)@disk catalyst prepared for RDE reference measurements. f-h) Scanning electron microscopic (SEM) inspection of the Cu-foam(30 s)@disk catalyst. i-j) Optical micrographs (side-view) of the Cu-foam (30 s) on the Cu disc embedded into the Teflon RDE holder. Figure S11. ECSA determination using the viologen method (see Experimental section and Figure S4). a-b) Data obtained for the Cu disc. c-d) Data obtained for the Cu-foam(30 s)@disc sample. Figure S12. a-b) Angular frequency (rotational speed) depending linear sweep voltammograms (LSVs) demonstrating nitrate mass transport limitations for both the nitrate→nitrite and the nitrate→ammonia transformation at pH 14 (100 mmol L -1 nitrate concentration). c-d) Corresponding RDE data for pH 14 using a nitrate concentration of 500 mmol L -1 . Obviously mass transport limitations are omitted at these higher nitrate concentrations. Figure S13. Representative chrono-amperometric data (total current density (TCDgeo) versus time plots) corresponding to Figure 5d. Note that the electrolysis performance data presented in Figure 5 is derived from three independent electrolyses per applied potential.    From Fick's first law follows that in case of a stationary diffusion layer thickness δ, the limiting current density can be calculated as where n denotes the number of electrons taking part in the reaction (n = 8 for nitrate→ammonia and n = 2 for nitrate→nitrite reduction); F = 96485.3 C mol -1 is Faraday's constant and D = 1.902 · 10 -5 cm 2 s -1 is the (approximate) diffusion coefficient 1 of nitrate ions. The relationship between the achievable (limiting) current density and the diffusion layer thickness is shown in Figure S17 for different nitrate concentrations and reaction routes (nitrite or ammonia formation). In general, the diffusion layer thickness δ can be decreased (the limiting current increased) by the application of agitation (stirring). Natural convection, that occurs by a slow motion of electrolyte caused by electrolysis-induced density gradients, usually establishes a few hundreds of micrometers thick diffusion layers. By means of explicit agitation (e.g., by rotating the electrode), δ can be decreased to the range of tens of micrometers, while intensive gas formation can result in only a few micrometers thick diffusion layers on planar electrodes. For porous surfaces with considerable surface roughness, under given stirring conditions, the achievable current density can, according to the literature, increase about 2-fold compared to what is shown in Figure S17. 2 According to Ibl and Venczel, 3,4,5 the continuous release of H2 bubbles of an average radius r from the electrode surface could establish a diffusion layer thickness where D denotes the diffusion coefficient of the reacting species (in our case, of nitrate ions), τ is the relative coverage of the surface by gas bubbles, and vg is a velocity term gained by normalizing the (volumetric) rate of gas evolution to the surface area of the electrode.
The experimentally observed limiting currents for "non-stirred" (natural convection only case) and gas evolution conditions presented in Figure 5 and 7 are well within the expected range indicated in Figure  S17. Figure S18. Product distribution of 30 min. electrolyses carried out at pH 14 (1 mol L -1 KOH) in 100 mmol L -1 KNO3 electrolyte solution. The product analysis was extended towards online gaschromatography exemplarily demonstrating that hydrogen is indeed the prevalent by-product of the electrolyses performed at potentials more negative than -0.4 V vs. RHE. The product distribution was derived from averaging three individual electrolyses. Note that for more cathodic electrolysis potentials the FE values do not sum up in all cases to 100%. This is most likely due to a partial loss of formed ammonia by diffusion into the anolyte (see Figure S21). The efficiencies for ammonia production may be in reality even higher than displayed in the FE plots. One important finding is that the anion exchange membrane is permeable for ammonia/ammonium (see Figure S21).  Figure S17. A geometric surface area of 1 cm 2 is realized by masking the wafer coupon with insulating Teflon tape. b-c) Scanning force microscopic inspection of the Cu wafer coupon sample. d-e) Voltammetric data (viologen method, see Experimental section) used for the ESCA determination. f) Voltammetric data demonstrating the nitrate reduction on the Cu wafer coupon sample. The potential sweep rate was 25 mV s -1 . Figure S20. a-d) Electrolysis data (30 minutes electrolysis duration) obtained for a planar Cu wafer coupon sample (see Figure S19). e-f) Electrolysis data (30 minutes electrolysis duration) obtained for the Cu-foam(30 s)@mesh sample.
The comparison confirms the effective trapping and further reaction of nitrite into ammonia in case of the Cu foam catalyst (see Figure S20c and Figure S20g). By contrast to that, nitrite intermediates are readily released into the electrolyte in case of the ideally planar Cu wafer coupon sample. Of note is that these trapping effects are only effective under "competitive" experimental conditions where both products, ammonia and nitrite, form on the Cu catalyst and nitrite can in principle be released into the electrolyte (e.g., at pH 14 and 0.1 mol L -1 nitrate concentrations at potentials above -0.4 V vs. RHE). At more cathodic electrolysis potentials, however, adsorbed nitrite directly reacts further into ammonia.
Trapping effects seem to be of less importance at these electrolysis conditions. In either case (Cu foam, wafer coupon) no nitrate could be detected as a by-product of the nitrate reduction at potentials more negative than -0.4 V vs. RHE. A comparison further confirms the appearance of pronounced limiting currents (plateaus in the TCDgeo vs. E plots) for the nitrate reduction in both cases before the onset of the HER. The limiting current in case of the porous Cu foam is higher by a factor of ca. 2 compared to the planar Cu wafer coupon (see discussion of Figure S17, reference 2). The observed "boost" of the ammonia partial current density sets in at less cathodic potentials in the case of the Cu foam catalyst and is much more pronounced than in case of the planar Cu wafer coupon sample (see Figure S20d and Figure S20h). This observation can be rationalized, at least partly, by an ECSA which is higher by a factor of ca. 7 in case of the porous Cu foam (30 s deposition, see table S1) when compared to the planar wafer coupon surface (ECSA ca. 1 cm 2 ).

Figure S21
. Time dependent accumulation of ammonia in the anolyte evidenced by the indophenol analysis approach. Samples were taken from the anolyte as a function of the electrolysis time during nitrate reduction at -0.7 V vs. RHE at pH 14 and 100 mmol L -1 KNO3 electrolyte solution. A precise quantification through UV-vis was not carried out in this case due to expected partial re-oxidation of ammonia at the anode. These experiments, however, clearly demonstrate that ammonia can pass the anion exchange membrane. The FE values for ammonia production derived from the analysis of the catholyte might be in reality even higher in particular following extended electrolyses at higher reaction rates.  Table S1. Characteristics of the Cu-foam@mesh catalysts depending on the Cu deposition time. The data presented herein correspond to Figure 3 and Figure S4.  Table S4. Nitrate concentration (10 mmol L -1 , 100 mmol L -1 , and 500 mmol L -1 ) and potential dependent Faradaic efficiency data for ammonia (main NO 3 − RR product) and nitrite (parasitic NO 3 − RR by-product).
The electrolysis data were obtained after 30 min of potentiostatic electrolysis at pH 14. For each electrolysis a newly prepared catalyst was used. The  Figure 7b. Table S6. Nitrate concentration (10 mmol L -1 , 100 mmol L -1 , and 500 mmol L -1 ) and potential dependent partial current densities (PCDs) for ammonia (main NO 3 − RR product) and nitrite (parasitic   Table S7. Nitrate concentration (10 mmol L -1 , 100 mmol L -1 , and 500 mmol L -1 ) and potential dependent partial current densities (PCDs) for ammonia (main NO 3 − RR product) and nitrite (parasitic NO 3 − RR byproduct). The electrolysis data was obtained after 30 min of potentiostatic electrolysis at pH 14. The current densities presented from Table S6 were normalized (herein) to the electrochemically active surface area (ECSA). Table S8. Nitrate concentration (10 mmol L -1 , 100 mmol L -1 , and 500 mmol L -1 ) and potential dependent Faradaic efficiency data for ammonia (main NO 3 − RR product) and nitrite (parasitic NO 3 − RR by-product).
The electrolysis data was obtained after 30 min of potentiostatic electrolysis at pH 7. For each electrolysis a newly prepared catalyst was used. The   Figure 7e. Table S10. Nitrate concentration (10 mmol L -1 , 100 mmol L -1 , and 500 mmol L -1 ) and potential dependent partial current densities (PCDs) for ammonia (main NO 3 − RR product) and nitrite (parasitic  Table S11. Nitrate concentration (10 mmol L -1 , 100 mmol L -1 , and 500 mmol L -1 ) and potential dependent partial current densities (PCDs) for ammonia (main NO 3 − RR product) and nitrite (parasitic   Table S12. Time dependent electrolysis data corresponding to Figure 8 a and c (potentiostatic electrolyses at -0.3 V vs. RHE). The current densities were normalized to the geometric surface area. The "initial" values (determined 30 min after start of the respective continuous electrolysis) are highlighted in red.  The presented data correspond to Figure S20.