Electrochemical Nitrogen Reduction Reaction on Noble Metal Catalysts in Proton and Hydroxide Exchange Membrane Electrolyzers

Five noble metal catalysts (Pt/C, Ir/C, Pd/C, Ru/C, and Au/C) are evaluated for electrochemical nitrogen reduction reaction (ENRR) to produce ammonia in both proton exchange membrane (PEM) and hydroxide exchange membrane (HEM) electrolyzers (PEMELs and HEMELs). The competing hydrogen evolution reaction (HER) is found to be the dominant reaction on all catalysts tested in both PEMELS and HEMELs, which is consistent with recent computational predictions that metallic catalysts are unlikely to be selective for ENRR. With the increase of applied potentials, the rate of HER increase signiﬁcantly, which suppresses the ENRR, leading to signiﬁcant decreases of faradaic efﬁciencies. Leaching of quaternary ammonium from HEM is found to interfere with ammonia quantiﬁcation, which necessitates a pretreatment protocol. Due to the relatively slow kinetics of HER in alkaline solution, faradaic efﬁciencies in HEMELs are generally higher than those in PEMELs. We believe that these results provide a solid baseline for future research on ENRR in both PEMELs and HEMELs. (PEMELs) and HEMELs. We establish a reliable method to evaluate the ammonia production rate for ENRR with a membrane electrode assembly (MEA) conﬁguration. We show that the competing hydro- gen evolution reaction (HER) is the dominant reaction on all noble metal catalysts in PEMELs and HEMELs. Ammonia production rates are higher in PEMELs by roughly one order of magnitude higher than those in HEMELs, however, the trend is reversed with faradaic efﬁciencies for ENRR.

Ammonia synthesis is one of the foundational chemical processes to the human society, which is estimated to supported approximately 27% of the world's population over the past century. 1 The development of the Haber-Bosch process revolutionized modern agriculture, 2 however, the centralized, energy and carbon intensive nature of the Haber-Bosch process [3][4][5] leaves many aspects to be desired. Distributed and modular ammonia synthesis via the electrochemical nitrogen reduction reaction (ENRR) at or close to ambient conditions, powered by renewable electricity, is an attractive alternative because it allows as-needed production of ammonia, and in turn N-fertilizers, from ubiquitously available resources, i.e., N 2 and water. 4 Widespread adoption of ENRR for ammonia production could drastically reduce the carbon footprint of agricultural activities. In addition, ENRR is compatible with the intermittency of renewable energy sources, e.g., solar and wind, as the ammonia and N-fertilizers can be produced and stored when renewable electricity is available.
Electrochemical fixation of atmospheric nitrogen was first attempted in 1908 even before the establishment of the Haber-Bosch process, to make nitric acid via electric discharge. 6 Further studies of ENRR have generally been focused on high temperature proton conductors at 500 • C, 7 however, the high operating temperature and the stability of ammonia at such conditions make it unsuitable for distributed deployment. Proton exchange membranes (PEMs) can achieve high proton conductivity (∼100 mS·cm −1 at 25 • C for Nafion) at ambient or slightly elevated temperatures, which opens the possibility for electrochemical ammonia synthesis at those mild conditions. 8 Although Nafion-based low temperature electrochemical ammonia synthesis has been demonstrated in several reports, [9][10][11][12][13][14][15] no systematic investigation of noble metal catalysts exists. 2 Further, the alkaline environment of hydroxide exchange membranes (HEMs) promises the employment of non-noble metal catalysts and no acid/base reaction between the produced ammonia and the membrane is expected. The slower HER kinetics in base also could lead to an improved selectivity toward ENRR. 16 However, ENRR in HEM-based electrolyzers (HEMELs) has not been extensively investigated. In particular, no reliable strategy for ammonia quantification in HEMELs has yet been developed, owing to the possible leaching of quaternary ammonium cations from HEMs (or ionomer) that could interfere with the commonly used Nessler's method. 17 In this work, we survey five noble metal catalysts (Pt/C, Ir/C, Pd/C, Ru/C, and Au/C) for ENRR in both PEM-based electrolyzers = These authors contributed equally to this work. z E-mail: yanys@udel.edu; bxu@udel.edu (PEMELs) and HEMELs. We establish a reliable method to evaluate the ammonia production rate for ENRR with a membrane electrode assembly (MEA) configuration. We show that the competing hydrogen evolution reaction (HER) is the dominant reaction on all noble metal catalysts in PEMELs and HEMELs. Ammonia production rates are higher in PEMELs by roughly one order of magnitude higher than those in HEMELs, however, the trend is reversed with faradaic efficiencies for ENRR.
Preparation of the working electrode and electrochemical measurements.-Electrochemical measurements were performed in a three-electrode cell with a saturated calomel electrode (SCE; Princeton Applied Research) immersed in a Luggin capillary (Princeton Applied Research) filled with 2 M KNO 3 as the reference electrode, a Pt wire (PINE Instrument) as the counter electrode, and a glassy carbon (5-mm diameter, PINE Instrument) as the working electrode. 18 Ink solutions of Pt/C, Ir/C, Pd/C, Ru/C, and Au/C were prepared by dispersing Pt/C, Ir/C, Pd/C, Ru/C, and Au/C in 0.05 wt% Nafion IPA solution followed by ultrasonication for 1 h. The thin-film electrodes were made by pipetting 2 μL of the ink solutions one, two, or four times onto pre-polished glassy carbon electrodes with final metal loadings of 2−20 μg metal /cm 2 disk . Cyclic voltammetry (CV) measurements for these noble metal catalysts were carried out at room temperature in Ar-saturated 0.1 M H 2 SO 4 and 0.1 M KOH solution at a scanning rate of 50 mV/s, respectively. All potentials were converted to values in reference to the reversible hydrogen electrode (RHE).

Preparation of MEA and ENRR tests.-
The ink solutions of noble metal catalysts were mixed into a mixture of H 2 O:IPA (50:50) with 30 wt% Nafion ionomer and then hand-sprayed onto a carbon paper to an approximate metal loading of 0.4 mg cm −2 to serve as the cathode. The ink solution of Pt/C was prepared in the same way and was subsequently sprayed onto a Nafion-211 membrane using a Sonotek sprayer to reach a final Pt loading of 0.4 mg cm −2 , which was used as the anode. The anode is fed with 1 atm of H 2 , which turns it into a RHE. Thus, the cell potential is numerically identical to the cathode potential vs. RHE and the ENRR can be performed at well-defined potentials. The membrane was then covered with a Sigracet 39 BC carbon paper anode gas diffusion layer and assembled with the cathode into an MEA with an active geometric cross-sectional area of 5 cm 2 . The production of ammonia was conducted with the cathode and anode fed by high-purity N 2 (99.999%) and H 2 (99.999%) under 1 atm at a constant flow rate of 0.1 L min −1 , respectively. The temperature of the cell is controlled to be 80 • C. The temperatures of the humidifiers for cathode and anode are controlled to be 85 • C to maintain 100% relative humidity in the electrolyzer. The ammonia produced on the cathode and anode were collected by 1 mM H 2 SO 4 . After the tests in a PEMEL, the membrane was soaked in 20 mL of 3 M H 2 SO 4 for 24 h to extract the ammonia inside the membrane. At least two separate MEAs were tested to observe the reproducibility of the results.
Quantification of ammonia.-The amount of ammonia was determined using Nessler's methods according to the following equation: Since the ammonia was collected in acid solution, KOH solution (3 M) was added to adjust the pH of the solution to ∼12.5. A calibration curve was made by using (NH 4 ) 2 SO 4 solutions with known concentrations using a photometer (HI 96715, HANNA). 19

Results and Discussion
Electrochemical surface areas (ECSAs) of as-purchased noble metal catalysts, i.e., Pt/C, Ir/C, Pd/C, and Au/C, were determined via CV scans performed in 0.1 M H 2 SO 4 and 0.1 M KOH ( Figure  1). For the commercial Pt/C and Ir/C catalysts, ECSAs were derived from charges associated with the desorption and/or adsorption of underpontential deposited (UPD) hydrogen and normalized to the corresponding metal loading on the working electrode with a charge density of 210 μC cm −2 . The ECSAs of commercial Pd/C and Au/C catalysts were determined from the reduction peaks of PdO and AuO with a double layer correction and a charge density of 424 μC cm −2 and 386 μC cm −2 , respectively. 18,20 Copper under-potential deposition (Cu UPD) was carried out in 2 mM CuSO 4 (in 0.1 M H 2 SO 4 ) to evaluate the ECSA of the Ru/C catalyst with a charge density of 420 μC cm −2 . 18 Table I summarizes the specific ECSAs of the PGM catalysts in both acidic and alkaline solutions, which were used to normalize ammonia production rates in PEMELs and HEMELs, respectively. ENRR rates were measured using an MEA configuration due to the low solubility of N 2 in aqueous electrolytes (Figure 2). It is worth noting that Pt/C at a high loading of 0.4 mg Pt /cm 2 was employed on the anode in 1 atm of H 2 in both, and thus the anode side functions as a RHE and the cell potential is numerically identical to the cathodic potential. For the ENRR in PEMEL, the first step was to conduct a break-in procedure for 1 h at the desired applied potential, in which the cathode site was fed with H 2 rather than N 2 , in order to activate the cathode catalyst. After the break-in, the cathode feed was switched to N 2 for the ENRR. The current responses for the different cathode materials were generally stable with time at both −0.  the catalysts in HEMELs, except for Ru, than those in PEMELs under otherwise identical conditions. Ru shows similar HER activity in acidic and alkaline environments, 21 which is the likely cause of the similar currents in PEMEL and HEMEL as HER is the dominant reaction in both cases. The catalytic activity and selectivity of noble metal catalysts for the production of ammonia were investigated at two cathodic potentials, i.e., −0.2 V and −0.4 V (Figure 3). At −0.2 V, all noble metal catalysts showed similar ammonia production rates, with Ir being the most active with an ammonia production rate of 2.09 × 10 −12 mol cm ECSA −2 s −1 . The ammonia production rates at −0.4 V were very close to those at −0.2 V on all noble metal catalysts. The highest ammonia production rate (2.12 × 10 −12 mol cm ECSA −2 s −1 ) was still observed on Ir. Ammonia production rates of Pt in this work are comparable to those reported by Kordali et al. but about one order of magnitude lower than the previously reported values by Tao and co-workers. 9,12 Methodological differences in the quantification of amounts of ammonia produced in this work from previous works could be a major contributor in the differences of measured rates. 9 We recommend testing multiple MEAs to improve reproducibility and MEAs that have not been pre-exchanged with ammonium whose leaking can significantly interfere with the quantification of ammonia production rate and faradaic efficiency. In the previous work, NH 4 +exchanged membranes were used for ENRR. 9 Since H + produced on the anode is transported via the membrane to the cathode, the possibility of the NH 4 + in the PEM prior to reaction getting displaced by H + and evolved on the hydrogen side cannot be excluded, especially when the amount of NH 4 + in the NH 4 + -exchanged membrane was comparable to the amount of detected ammonia. The fact that ammonia was also observed on the anode side in the previous study indicates that leaching of ammonium, either through ion-exchange or degradation, from the Nafion membrane is likely occurring. 9 The amount of ammonia within the membrane after testing was not demonstrated conclusively. Therefore, we conclude that quantifying amounts of ammonia in Nafion and evolved at the cathode in this work is a more accurate and reliable method. No detectable level of ammonia was observed at the effluent of anode in any of the tests. In order to extract the ammonia inside the membrane, the membranes after reaction were soaked in 3 M H 2 SO 4 for 24 h in PEMELs in the current study. At both −0.2 V and −0.4 V, Ru showed the highest faradaic efficiencies (FEs), which were 0.053% and 0.015%, respectively. Notably, with the increased potential, the ammonia production rate remained essentially unchanged, but the FE decreased on average by roughly one order of magnitude, which indicates that the competing HER is more favorable at low potentials. Our results are consistent with the computational predictions that noble metals and metal alloys are unlikely to be selective for ENRR, which is attributed to the correlated binding energies of various surface intermediates in ENRR and the competing HER, known as the linear scaling relations (LSR). 2 According to previous experimental and computational results, conducting ENRR in an alkaline environment could potentially be more selective toward ENRR due to the slower kinetics of the competing HER on the cathode. 18,22,23 Inspired by this, we also investigated the ENRR performance of noble metal catalysts with HEMELs, in which commercial anion exchange membrane and ionomer were employed. Since the possible leaching of quaternary ammonium cations from the membrane and ionomer, which are the typical charge carrying groups for HEMs, could react with the Nessler reagent and interfere the quantification of ammonia from ENRR, it is critical to establish a reliable experimental protocol to separate contributions from ENRR and leaching of HEM and ionomer to the Nessler's results. Further, it is important to investigate the stability of membrane and ionomer under operating conditions to obtain accurate ammonia production rates. To test the possibility of leached quaternary ammonium cations from the commercial HEM and ionomer interfering with ammonia quantification, we conducted a control experiment in which both the cathode and anode were fed with H 2 at −0.2 V for 4 h, and thus the only source of nitrogen is from the membrane and ionomer, rather than ENRR. The exhaust solutions were collected to quantify the amount of quaternary ammonium leached from the membrane and/or the ionomer. It was found that there was a large amount of leached quaternary ammonium in both cathode and anode exhaust during the first hour. The total amount of leached quaternary ammonium (2.3 × 10 −6 mol) was around the same for all noble metal catalysts. No detectable amount of quaternary ammonium was observed in the exhaust afterwards, indicating that the leaching from membrane and ionomer after 1 h at −0.2 V is negligible. To further identify the distributions of leached quaternary ammonium, we conducted another control experiment in which no ionomer was used, and thus the only source of nitrogen is from the membrane. It was found that the total amount of leached quaternary ammonium was about 0.8 × 10 −6 mol, indicating that about two thirds of leached quaternary ammonium came from the ionomer, which was about half of the total amount of ionomer employed. Therefore, a pretreatment step with both cathode and anode fed with H 2 at −0.2 V for 1 h was added to all ENRR testing in HEMELs to ensure all detected ammonia comes from ENRR. The cathode feed was switched from H 2 to N 2 after 1h, and all ammonia detected afterwards was considered to be produced from ENRR. Ammonia was only detected in the effluent on the cathode side, but not on the anode side, over all noble metal catalysts in HEMELs, which indicates that the ammonia detected after the pretreatment is indeed from ENRR. As shown in Figure 4, ammonia production rates of all noble metal catalysts are close in alkaline environment, and are approximately one order of magnitude lower than those in acid. Again, Ir shows the highest ammonia production rate at −0.2 V in a HEMELs at 4.3 × 10 −13 mol cm ECSA −2 s −1 . Interestingly, due to the slower kinetics of HER in basic medium, most noble metal catalysts show higher FEs in HEMELs than in PEMELs. Au showed the highest FE of 0.55% in HEMELs, which is about one order of magnitude higher than that in PEMELs, indicating that Au might be a good candidate catalyst for ENRR in HEMELs. Recent works reported high faradaic efficiencies for Au sub-nanoclusters (8.11% 2 ) and tetrahexahedral Au nanorods (4.02%), which were attributed to their unique nanostructures. 24,25 In addition to their catalytic activity and selectivity, we also investigated the durability of commercial Pt/C catalyst for ENRR in both PEMELs and HEMELs. As shown in Figure 5, the amounts of ammonia produced in the first 0.5 h are around 50% of those produced within 1 h, which indicates that the ammonia production rate in 1 h are constant in both PEMELs and HEMELs. X-ray photoelectron spectroscopy (XPS) was utilized for the characterization of fresh and spent Pt/C catalysts employed in PEMELs ( Figure 6). The binding energies of Pt 4f 7/2 and Pt 4f 5/2 in the XPS spectra of both fresh and spent Pt/C catalysts were 71.4 and 74.6 eV, respectively, which are similar to the binding energies of previously reported Pt-based catalysts. 26 The XPS results suggest that only metallic Pt are present in the fresh and spent Pt/C catalysts in PEMELs, and thus the surface state of commercial Pt/C catalyst does not change to any detectable level after ENRR for 1 h.

Conclusions
In conclusion, we have investigated the catalytic performance of five noble metal catalysts (Pt/C, Ir/C, Pd/C, Ru/C, and Au/C) in ENRR at well-defined cathode potentials (−0.2 and −0.4 V vs. RHE) in both acidic and alkaline environments. HER is the dominant reaction on all catalysts in both PEMELs and HEMELs, with ENRR faradaic efficiencies well below 1%, which is consistent with previous computational predictions. Ammonia production rates on noble metal catalysts in PEMELs is roughly one order of magnitude higher than those in HEMELs, while the faradaic efficiency for ENRR is higher in HEMELs than PEMELs. The low ENRR FEs show that HER is the preferred reaction on these noble metal catalysts in both acid and  base, likely due to the LSR between the hydrogen binding energy and the nitrogen binding energy. Thus, only catalysts with decoupled hydrogen and nitrogen binding energies can be selective toward ENRR. This is consistent with the retarded HER kinetics in alkaline media compared to acidic media. We show that the leaching of quaternary ammonium cations from HEM and ionomer interfere with the ammonia quantification using the Nessler's method, and a pretreatment step is necessary to avoid any ambiguity in the quantification of ammonia produced in ENRR. Time-dependent results show that Pt/C is stable under operating conditions in both PEMEL and HEMEL within 1h. These results provide a solid baseline for future ENRR and ammonia production studies using MEA technique.