Stabilizing calcium nitride for efficient, long-term electrochemical ammonia synthesis

Significance

Electrochemical synthesis of ammonia from nitrogen under ambient conditions is crucial for sustainable fertilizer production but faces challenges due to unstable intermediates and poor long-term performance. This work addresses fundamental limitations by stabilizing calcium nitride intermediates through a precise selection of solvents, electrodes, and cell configurations. We provide mechanistic insights into calcium-mediated nitrogen activation, enabling stable, selective ammonia synthesis at higher current densities and extended operation times. These findings set foundational design principles for robust, high-performance electrochemical ammonia production, bringing this sustainable technology significantly closer to industrial viability.

Abstract

Electrochemical ammonia synthesis at ambient conditions via calcium-mediated nitrogen fixation holds considerable promise but is impeded by fundamental gaps such as poor gas–liquid interface stability, sluggish hydrogen oxidation reaction (HOR) kinetics, and instability of the critical intermediate calcium nitride. To systematically address these barriers, we i) introduced a high surface-area Ni-based anode specifically selected for enhancing HOR kinetics and minimizing solvent oxidation; ii) substituted the conventionally used tetrahydrofuran solvent with dimethoxyethane (DME) to significantly improve chemical stability; and iii) developed a tailored flow cell configuration to enhance gas–liquid mass transport and stabilize reaction intermediates. Employing in-situ Raman spectroscopy and X-ray photoelectron spectroscopy, we provided direct evidence of stabilized calcium nitride formation, elucidating the crucial roles of solvent stability and electrode composition in sustaining reactive intermediates. As a result of these combined innovations, our system demonstrates substantial performance improvements, achieving a Faradaic efficiency (FE) of 34.35 ± 1.76% in short-term tests and sustaining ~20% FE over extended continuous operation (~56 h). At elevated current densities, the improved gas–liquid interface stability enables robust ammonia production, reaching partial current densities of approximately 219 mA cm⁻² at ~29% FE. Isotope-labeling studies with ¹⁵N₂ confirmed the direct electroreduction of N₂, while kinetic analyses underscored the impact of anode material selection on HOR efficiency and overall electrochemical stability. These insights establish critical mechanistic understanding and clear design principles for future calcium-mediated electrochemical nitrogen fixation systems, enabling stable, efficient, and selective ammonia synthesis.

Ammonia is a crucial commodity chemical essential for sustaining life, and it is also emerging as one of the best carbon-free fuels. To meet the ever-increasing demand for ammonia, the Haber–Bosch process is employed for large-scale production. However, this process is highly energy-intensive, requiring extremely high temperatures and pressures, and it is responsible for approximately 1.3% of global CO₂ emissions (1–5). Therefore, there is an urgent need to develop technologies that can facilitate green ammonia production.

Electrochemical ammonia synthesis directly from nitrogen and water, powered by renewable energy, presents a promising alternative. This method could enable distributed ammonia production in smaller-scale devices, reducing reliance on large, centralized plants. Significant research has been conducted on lithium-mediated ammonia synthesis (Li-MAS), showing substantial progress in achieving high Faradaic Efficiency (FE) and current density (6–26). However, most Li-MAS investigations have been performed in batch-type electrochemical cells, such as one-compartment cells and autoclaves, where nitrogen must be dissolved in the electrolyte to participate in the reaction, even at high N₂ pressures (e.g., 5 to 100 bar). This configuration is mass transport-limited with respect to nitrogen due to the low solubility of N₂ in nonaqueous electrolytes. Moreover, most Li-MAS (7, 9, 12, 15–23, 27, 28) studies face the limitation of using a sacrificial solvent as a proton donor and encounter difficulties in scaling up production in batch reactors.

Continuous ammonia synthesis at 1 bar N₂ pressure faces several critical challenges that hinder its practicality. These include low FE and poor energy efficiency, primarily due to mass transfer limitations. Additionally, high cell voltage requirements further impact overall performance. Another major concern is the use of sacrificial solvents as a proton source, which is unsustainable for long-term applications (7). Introducing a hydrogen oxidation reaction (HOR) on the anode, with hydrogen obtained from water splitting, is a promising approach that holds the potential to mimic the Haber–Bosch process. Suryanto et al. (28) attempted to resolve the issue of sacrificial solvents by introducing a phosphonium proton shuttle, conducting their experiments at 0.5 bar H₂ and 19.5 bar N₂ in an autoclave. However, in batch cells, the mass transfer limitation of H₂ remains a significant challenge (7). Additionally, H₂ can react with metallic lithium to form lithium hydride, which hinders the ability to activate nitrogen at room temperature (7). To circumvent these transport limitations, Lazouski et al. (12) proposed the use of a stainless-steel cloth (SSC) as the gas diffusion electrode (GDE) in a three-compartment cell, achieving an ammonia FE of 35%. However, this system was only stable for 5 to 8 min at a current density of 20 mA/cm².

Despite these advancements, Li-MAS remains hindered by poor energy efficiency, which is notably lower than that of the conventional Haber–Bosch process. This inefficiency primarily stems from lithium’s highly reducing electroplating potential (~-3.04 V vs. SHE). Additionally, sustaining Li-MAS at high current densities is challenging due to electrolyte degradation, which disrupts lithium recovery, while the high cost of lithium salts further limits economic feasibility (10, 25). Choosing a metal with a less reducing electroplating potential could significantly improve energy efficiency. These challenges highlight the necessity of exploring alternative metal-mediated systems that offer greater stability and cost-effectiveness. This study motivates the exploration of alternative metal mediators such as magnesium, barium, strontium, and sodium for nitrogen reduction. These choices are supported by previously identified design criteria for effective mediators, including the ability to form nitrides spontaneously in the presence of N₂, maintain stable nitrogen vacancies, and exhibit favorable N₂ binding and activation kinetics. For example, magnesium and calcium can readily form stable nitrides (Mg₃N₂ and Ca₃N₂) under relatively mild conditions, and recent theoretical studies suggest their lower plating potentials compared to lithium could enhance overall energy efficiency. Barium and strontium also form stable nitrides (Ba₃N₂ and Sr₃N₂) and are notable for their high nitrogen diffusivity within the bulk nitride lattice, potentially supporting continuous nitrogen incorporation and release. Sodium, while its nitride (Na₃N) is less thermodynamically stable, has gained interest due to its abundance and capacity for fast nitridation under certain conditions. Recent computational and experimental work (29–39) has examined these metals as potential mediators, highlighting how differences in nitride stability, redox potential, and salt solubility could open pathways toward more scalable and efficient electrochemical ammonia synthesis beyond lithium.

In our previous work, we identified calcium and magnesium as potential mediators for ammonia synthesis (25, 29–37, 39). We also conducted calcium-mediated ammonia synthesis at 6 bar and achieved an impressive FE of 49%. Moreover, recent works by Fu et al. (4, 7, 25) have developed stable continuous systems using a platinum-gold alloy as the anode, demonstrating the feasibility of continuous LiMAS (7) and CaMAS (25) approach.

In this study, we address the prevailing challenges in the emerging continuous calcium-mediated nitrogen reduction, including long-term operational stability, stable economic HOR electrodes, nitrogen utilization efficiency, and robust cell materials resistant to corrosion in harsh organic electrolytes (Scheme 1). Additionally, we incorporate an ether-based solvent with superior stability compared to conventional tetrahydrofuran, mitigating electrolyte degradation under high potentials often reported in prior literature (40). By implementing a continuous calcium-mediated approach in a flow cell, we systematically evaluated anode stability using voltage and kinetic studies to identify an economically viable electrode. Initial testing in a commercial electrolyzer setup revealed that a 350 nm Pt-coated stainless steel mesh anode exhibited consistent performance, sustaining an FE of 9.39 ± 1.13% over four consecutive runs at lower applied potentials. To overcome limitations such as poor electrical contact, inefficient liquid–gas contact, electrolyte seepage through GDEs, and intricate cell assembly, we developed a custom air-tight continuous PEEK electrolyzer design employing Ni foam as both anode and cathode. This configuration significantly enhanced performance, achieving an ammonia FE of 34.35 ± 1.76%. Long-term and high-current density studies were conducted to evaluate system performance under industrially relevant conditions. The system demonstrated stable operation over 56 h with an ammonia selectivity of approximately 20%. Under high current conditions, it achieved an ammonia partial current density of ~219 mA cm⁻² and an FE of ~29%, highlighting its strong potential for scalable electrochemical ammonia production. We achieved a per-pass conversion efficiency of nearly 5% for high current density operation, which significantly surpasses values reported in previous studies. This highlights the enhanced efficiency and practical applicability of our system. Isotope labeling and control experiments confirmed N₂ as the only nitrogen source, while XPS and SEM-EDS analyses verified metallic Ca and N presence, strongly indicating Ca₃N₂ formation. Further validation through in-situ Raman spectrometry provided direct evidence of calcium nitride generation, reinforcing the mechanistic insights of our system.

Scheme 1.

Schematic illustration of advancements in the calcium-mediated ammonia synthesis system, transitioning from a commercially available electrolyzer to our optimized PEEK flow electrolyzer. The new setup enhances system stability, enables the use of more stable and cost-effective anodes, and improves calcium plating and cell stability. In this system, Ca²⁺ ions are reduced at a Ni foam cathode to deposit metallic Ca⁰, which reacts with N₂ to form Ca₃N₂. The solid electrolyte interphase (SEI) plays a key role in facilitating nitrogen diffusion to the Ca⁰ layer. Subsequent stepwise protonation of Ca₃N₂ by ethanol (EtOH) yields ammonia, while protons are continuously regenerated via hydrogen oxidation at the anode.

Mechanism

In Li-MAS, lithium metal is electrochemically deposited onto the cathode, where it reacts with molecular nitrogen to form lithium nitride. This intermediate subsequently undergoes protonation, yielding ammonia. We hypothesize that Ca-mediated ammonia synthesis occurs in a process analogous to Li-mediated ammonia synthesis (10), i.e., based on the following reaction scheme:

Calcium Electrodeposition:

$${\mathrm{Ca}}^{2+}+2{\mathrm{e}}^{-}\to \mathrm{Ca}(\mathrm{s}),$$ [1]

Calcium Nitridation:

$$6\mathrm{Ca}(\mathrm{s})+2{\mathrm{N}}_2(\mathrm{aq}.)\to 2{\mathrm{Ca}}_3{\mathrm{N}}_2(\mathrm{s}),$$ [2]

Calcium Nitride Protonation:

$${\mathrm{Ca}}_3{\mathrm{N}}_2(\mathrm{s})+\mathrm{HX}(\mathrm{aq}.)+{\mathrm{e}}^{-}\to {\mathrm{Ca}}_3\mathrm{N}-\mathrm{NH}+{\mathrm{X}}^{-},$$ [3]

$${\mathrm{Ca}}_3\mathrm{N}-\mathrm{NH}+\mathrm{HX}(\mathrm{aq}.)+{\mathrm{e}}^{-}\to {\mathrm{Ca}}_3\mathrm{N}-{\mathrm{NH}}_2+{\mathrm{X}}^{-},$$ [4]

$${\mathrm{Ca}}_3\mathrm{N}-{\mathrm{NH}}_2+\mathrm{HX}(\mathrm{aq}.)+{\mathrm{e}}^{-}\to {\mathrm{Ca}}_3\mathrm{N}-{\mathrm{NH}}_3+{\mathrm{X}}^{-},$$ [5]

Ammonia Desorption:

$${\mathrm{Ca}}_3\mathrm{N}-{\mathrm{NH}}_3\to {\mathrm{Ca}}_3\mathrm{N}+{\mathrm{NH}}_3(\mathrm{aq}.).$$ [6]

The process begins with the electrodeposition of Ca on a substrate by dissolving calcium salt in a nonaqueous solvent and applying a strongly reducing bias to the cathode (Calcium Electrodeposition, Eq. 1). The deposited Ca metal spontaneously reacts with N₂, either dissolved or gaseous in a gas diffusion electrode setup, to form calcium nitride (Ca₃N₂) (Calcium Nitridation, Eq. 2). Subsequently, calcium nitride undergoes protonation using a proton donor such as alcohol (HX) (Eqs. 3–5), facilitating its conversion into ammonia, which remains adsorbed on the Ca₃N₂ surface (Calcium Nitride Protonation). The ammonia then desorbs, creating a nitrogen-vacancy in the calcium nitride (Eq. 6). To clearly interpret the nitrogen-containing intermediate formed during the reaction, we can consider calcium nitride (Ca₃N₂) as Ca_1.5N, which effectively delivers one equivalent of ammonia while leaving reactive calcium sites on the surface. These remaining sites may follow two possible pathways: 1) reaction with ethanol to form calcium ethoxide, or 2) reactivation of N₂ via a Mars–van Krevelen-type mechanism, leading to further ammonia generation. While our current experimental data primarily confirm the formation of ammonia from the initial reduction step, these subsequent transformations represent promising routes toward establishing a catalytic cycle. The overall process proceeds predominantly through an associative mechanism (41, 42). In this mechanism, N₂ molecules adsorb at reactive surface nitrogen vacancies in the metal nitride (e.g., Ca₃N₂), and are subsequently protonated and reduced via stepwise proton-coupled electron transfer (CPET) steps to yield NH₃. This avoids the need for full N₂ dissociation into separate adsorbed nitrogen atoms prior to hydrogenation, which would correspond to a purely dissociative mechanism. Scheme 1 presents the reaction scheme for the continuous CaMAS system, depicting the cathodic mechanism of calcium-mediated ammonia synthesis (CaMAS) and ammonia formation. It also illustrates the HOR at the anode, where the generated H⁺ replenishes the ethanol proton shuttle, preventing its depletion.

To gain insight into the progression of Ca₃N₂ following protonation, two mechanistic pathways are hypothesized: i) further protonation of Ca–N species leading to additional ammonia formation, and ii) generation of nitrogen vacancies within the Ca–N lattice that may be replenished by gaseous N₂, restoring nitride species. Operando Raman measurements appear consistent with both possibilities. After deposition and nitridation, a calcium nitride peak is observed, which decreases upon ethanol introduction, suggesting protonation of Ca–N bonds. When N₂ is subsequently reintroduced, an increase in the calcium nitride signal is detected without additional calcium deposition, indicating either of the above possibilities. A second protonation step again reduces the signal, implying that the regenerated phase remains reactive. These observations suggest that Ca₃N₂ may undergo sequential protonation and nitridation steps rather than being irreversibly consumed (SI Appendix, Fig. S36).

This proposed mechanism differs slightly from those suggested in recent works (12, 15) for the analogous Li-mediated ammonia synthesis process. For Li-mediated processes, it has been proposed that lithium dissolves to form Li⁺, but we believe that concomitant dissolution does not occur under the strongly reducing conditions typically used, as the corrosion subreaction would be endergonic. However the precise mechanism remains poorly understood and is likely influenced by specific experimental conditions such as dynamic potential control, electrolyte and salt choice, and N₂ pressure. This study provides proof of concept for continuous Ca-mediated ammonia synthesis.

In the case of Ca-mediated ammonia synthesis, the N₂ molecule may initially be adsorbed onto the Ca surface. Ca, with its highly negative reduction potential and strong reducing power could help activate the N₂ molecule by weakening its triple bond (N≡N). This step is similar to the adsorption step in the Eley-Rideal mechanism, where nitrogen is positioned on the Ca surface. Hydrogen molecules in the bulk phase would then collide with the adsorbed nitrogen on the Ca surface. Because the nitrogen is activated, possibly through the formation of a Ca–N intermediate, the hydrogen can more readily react with the nitrogen to form ammonia. The final product, ammonia, is then released from the substrate surface after the reaction. The process is consistent with the Eley-Rideal mechanism, where only one reactant (N₂) is adsorbed on the surface, and the other reactant (H₂) is in the gas phase and directly interacts with the adsorbed species. Ca may play a key role in the activation and stabilization of nitrogen, which is required for efficient nitrogen reduction. Calcium’s interaction with nitrogen could be crucial for the direct reaction with hydrogen in an Eley-Rideal-like fashion. The Eley-Rideal mechanism is commonly observed in surface-catalyzed reactions, and the calcium-based catalyst for ammonia synthesis involves a surface interaction between the catalyst and reactants. Thus, the Eley-Rideal mechanism offers a plausible framework for understanding how nitrogen and hydrogen interact on the catalyst surface.

Results and Discussion

Continuous-Flow Commercial Electrolyzer Setup.

Building on our previous work on lithium (10) and calcium-mediated (39) ammonia synthesis in batch electrochemical cells, we found that controlling the proton donor concentration (typically ethanol) is critical. An optimal concentration is needed to promote the protolysis of lithium nitride (Li₃N) to produce ammonia while minimizing the unwanted parasitic reaction of metallic lithium with protons that leads to hydrogen evolution. Lazouski et al. (15) systematically varied ethanol concentration in tetrahydrofuran electrolytes and found that ammonia FE peaked at around 0.1 M ethanol; too little ethanol limited the protonation of Li₃N, while excess ethanol accelerated undesired lithium corrosion and hydrogen production (15, 43). Kinetic modeling by the same group further rationalized these observations by showing that ammonia selectivity is governed by differences in reaction orders between the nitridation and protonolysis steps (15, 43). Complementary in situ studies confirmed that even small deviations from the optimal proton concentration can significantly alter lithium speciation at the electrode, affecting ammonia production rates (44). These prior works collectively establish that precise control of proton donor concentration is critical: it must be high enough to efficiently convert Li₃N to NH₃, yet low enough to suppress direct lithium corrosion. Our study builds on these insights by exploring a comparable optimization window, corroborating that ammonia selectivity follows a similar nonlinear trend with proton donor concentration, and reinforcing the generality of this kinetic balance across different systems. For the continuous calcium-mediated ammonia synthesis in this study, we used 0.065 M EtOH as a proton shuttle. A stainless steel mesh, sputter-coated with 350 nm of platinum (Pt) target, served as the anode, while nickel foam was used as the cathode. It was also observed in our previous Li-MAS study that very high salt concentrations increase the viscosity of the solution, leading to higher resistance (10). This forces the system to operate at higher voltages, ultimately decreasing the overall energy efficiency of the process. Therefore, we employed 0.5 M Ca(ClO₄)₂.4H₂O dissolved in dimethoxy ethane as the electrolyte. The optimum concentration of this salt was identified in previous experiments involving Ca-MAS conducted in our lab (41). All the details regarding the chemicals used in this study are comprehensively outlined in the Supplementary Information for further reference.

In systems where gaseous reactants like N₂ at the cathode and H₂ at the anode are involved, GDEs are commonly used to address the challenge of low gas solubility in organic solvents and to reduce the resistance caused due to gas bubbles between the electrodes. Carbon-based GDEs with a hydrophobic layer are especially advantageous as they help prevent aqueous electrolytes from flooding the electrode (12). However, in nonaqueous electrolytes like DMF, the hydrophobic interaction in carbon-based GDEs is weakened, leading to electrolyte infiltration and flooding within the fibrous structure. This results in an increased diffusion path for gas molecules and a reduction in achievable current density. To mitigate this issue, we implemented a backpressure strategy, as demonstrated by Lazouski et al. (12), to regulate electrolyte penetration and wetting in GDEs when using nonaqueous systems. By optimizing material wetting properties and electrolyte flow dynamics, we effectively replicated GDE-like behavior, ensuring efficient gas–liquid interaction and sustaining high current densities. Expanding on these findings, we used a continuous flow electrolyzer, modeled after commercially available CO₂ electrolyzers designed to handle organic solvents. This electrolyzer consists of a metal end plate, a current collector plate, and graphite plates featuring gas flow fields for cathode and anode sides with serpentine channels to enhance gas residence time, as shown in SI Appendix, Fig. S1. However, this flow cell design lacked a dedicated pathway for electrolyte flow. We custom-designed a PEEK(polyether ether ketone)-based electrolyte flow channel to accommodate our organic electrolyte, ensuring efficient electrolyte distribution and system compatibility. This electrolyzer will be referred to as the continuous commercial electrolyzer setup throughout the remainder of the manuscript. N₂ is introduced at the cathode, while H₂ is supplied at the anode, serving as a proton source.

The selection of appropriate solvents is paramount for achieving stable and efficient electrochemical ammonia synthesis. While tetrahydrofuran (THF) has been commonly employed in lithium-mediated nitrogen reduction (Li-NRR) systems, its limitations, such as propensity for polymerization and relatively high volatility, can impede long-term operation. Recent advancements, as reported by Li et al., have highlighted the advantages of chain-ether-based electrolytes, specifically dimethoxyethane, for sustained ammonia electrosynthesis (45). Their work demonstrated that DME’s nonpolymerizing nature, higher boiling point (162 °C), and ability to form a compact and stable SEI significantly contribute to enhanced electrolyte stability and facilitate efficient ammonia release, enabling over 300 ho of continuous operation in a lithium-mediated system. Building upon these critical insights into solvent engineering, this study similarly leverages the benefits of DME in a calcium-mediated nitrogen reduction (Ca-NRR) system. The inherent stability and favorable electrochemical properties of DME are crucial for achieving the long-term operational stability and high performance observed in our calcium-mediated process. Our work presents several distinct advantages and contributions in comparison to prior research, including the aforementioned lithium-mediated systems. While previous studies have predominantly focused on lithium, our research successfully extends the application of stable chain-ether solvents to a calcium-mediated platform. Calcium, being an earth-abundant and cost-effective alternative to lithium, offers a promising avenue for scalable electrochemical ammonia synthesis. Demonstrating enhanced system stability and performance with DME in a Ca-NRR system broadens the scope of viable mediator chemistries. We show that the optimized electrode, solvent, and reactor configurations, particularly the use of DME, are instrumental in achieving high system stability and sustained ammonia production within a calcium-mediated framework. This underscores the transferable nature of beneficial solvent properties to alternative mediator systems, showcasing the robustness of our overall system design. By employing DME, our calcium-mediated system not only aligns with the principles of stable solvent selection identified in state-of-the-art lithium-mediated NRR but also advances the field by demonstrating comparable or superior performance within an earth-abundant electrochemical platform.

The flow rates of H₂ and N₂ were maintained at 75 sccm, and back pressure was applied at the gas outlets to prevent electrolyte ingress into the gas channels. The back pressure was maintained using a 10.5 cm high liquid column, with the gas outlets fully immersed at the bottom, as shown in SI Appendix, Fig. S11. The liquid column consisted of a 0.1 M H₂SO₄ acid trap. The cathode consists of a 3 mm thick commercially available nickel foam electrode, pressed against the PEEK block channel to maximize electrolyte contact. In the continuous commercial electrolyzer setup, electrical contacts (plates) are directly connected to the back of the nickel foam, ensuring uniform current distribution, as shown in Fig. 1A. During operation, calcium from the electrolyte is electrodeposited onto the nickel foam cathode, where it reacts with N₂ to form calcium nitride. This nitride then reacts with protons, generated through H₂ oxidation at the anode, to produce ammonia.

Fig. 1.

(A) Schematic and configuration of the continuous-flow reactor for electrochemical ammonia synthesis. (B–D) 30 SS mesh: Total Cell Voltage vs. time profiles for various electrode materials at a current density of −1 mA/cm². (B) 30 SS Mesh, (C) 350 nm Pt coated 30 SS Mesh, (D) 350 nm Ni coated 30 SS Mesh. (E–G) 80 SS Mesh: Voltage vs. time profiles for various electrode materials at a current density of −1 mA/cm². (E) 80 SS Mesh, (F) 350 nm Pt coated 80 SS Mesh, (G) 350 nm Ni coated 80 SS Mesh (Operating conditions: continuous commercial electrolyzer, electrolyte – 0.5 M Ca(ClO₄)₂ + 0.065 M EtOH in DME; 75 sccm N₂ and H₂ pressure, Working Electrode: Ni foam cathode and Counterelectrode: 350 Pt coated stainless steel mesh anode). The operating time varies for each electrode as the reaction ceases once the potentiostat reaches its potential limit of −48 V, leading to instrument overload.

A major challenge in optimizing the hydrogen oxidation (H₂ → H⁺) substrate was ensuring its stability, as finding a material that could sustain the reaction for extended periods was difficult. As reported by Fu et al. (7), the conventional catalyst, platinum, lacks stability for hydrogen oxidation in organic electrolytes due to the blockage of active sites by organic molecules or reaction intermediates. However, incorporating a platinum-gold alloy has been found to lower the anodic potential while preventing the undesirable decomposition of the organic electrolyte by suppressing the adsorption of organic poison species by Au. Another key factor was reusability, meaning the catalyst should deliver consistent performance over multiple cycles without needing replacement. In order to identify electrode materials that are simple, easily accessible, and straightforward to synthesize without adding complexity, we systematically evaluated different substrates based on their anode performance and potential requirements.

We systematically investigated stainless steel meshes with varying mesh sizes to optimize anode performance. The 20 SS mesh, characterized by its larger pore size, was unable to maintain adequate back pressure, resulting in frequent electrolyte seepage and flooding. Additionally, it exhibited significantly high voltage requirements, rendering it unsuitable for stable operation (SI Appendix, Fig. S14A). We evaluated platinized titanium mesh (~25 mesh) to improve performance, leveraging platinum’s known catalytic activity (7) for the HOR. However, this mesh also failed to ensure system stability, likely due to its relatively large pores, which led to excessive flooding and leakage (SI Appendix, Fig. S14B). Subsequently, we tested 30 SS mesh, which featured a finer structure. Despite a modest improvement, leakage persisted, and the system continued to operate at inefficient voltage levels (Fig. 1B). To address this, we sputter-coated platinum and nickel onto the 30 SS mesh. These modifications improved HOR activity and showed better electrochemical performance, yet voltage remained high, and issues such as flooding and gas crossover were still observed (Fig. 1 C and D). Further optimization using 80 SS mesh, with an even finer mesh size, led to improved back pressure retention and a notable reduction in electrolyte leakage. Nevertheless, voltage inefficiencies were still present (Fig. 1E). Nickel and platinum were further sputter-coated onto the 80 SS meshes to assess their impact on voltage–time performance. While both coatings significantly reduced voltage losses, the system experienced cell instability and electrolyte seepage, which affected overall performance (Fig. 1 F and G). The operating time varies for each electrode as the reaction ceases once the potentiostat reaches its potential limit of −48 V, leading to instrument overload. A sudden increase in cell potential is consistently observed after a certain operating time, indicating instability in the system. This instability is strongly influenced by the mesh size of the electrode. As the reaction proceeds, gases; mainly unreacted reactants and some products accumulate within the electrode structure. In electrodes with larger mesh openings (lower mesh numbers), the trapped gas causes a pressure buildup inside the cell. Due to the poor separation between gas and liquid in such configurations, both phases are expelled together through the gas outlet. This leads to significant electrolyte loss, disrupting the liquid–gas interface and causing the potential to rise sharply until the system reaches the voltage limit. In contrast, finer meshes (higher mesh numbers) allow for more stable operation by better managing gas release and maintaining phase separation for relatively longer times.

As the voltage data lacked kinetic insights, Tafel slope analysis of various materials, including bare, platinum-coated, and nickel-coated stainless steel meshes and a Pt-coated titanium mesh, was performed, as shown in SI Appendix, Fig. S14. Tafel analysis provides insight into the kinetics of HOR by quantifying the overpotential required to achieve a 10-fold increase in current density, expressed in units of V/dec. A lower Tafel slope indicates faster kinetics and lower overpotential losses, making it a critical metric for evaluating electrode efficiency. Among the tested materials, Pt-coated 80 SS mesh (0.199 V/dec, Fig. 2E) exhibited the lowest Tafel slope, followed by Ni-coated 80 SS mesh (0.211 V/dec, Fig. 2F), Pt-coated 30 SS mesh (0.26 V/dec, Fig. 2B), Ni-coated 30 SS mesh (0.32 V/dec, Fig. 2C), and platinized ~25 Ti mesh (0.345 V/dec, SI Appendix, Fig. S14D). Bare stainless steel meshes, particularly those with larger pore sizes, demonstrated considerably higher slopes (80 SS: 0.407 V/dec, Fig. 2D; 30 SS: 0.504 V/dec, Fig. 2A; 20 SS: 0.55 V/dec, SI Appendix, Fig. S14C), reflecting sluggish kinetics and poor electrocatalytic performance. These elevated Tafel slopes are notably higher than those typically observed in aqueous systems and reflect the constraints of a nonaqueous, aprotic electrolyte. In such media, although hydrogen gas is continuously supplied at the anode, its oxidation can be hindered by factors commonly observed in nonaqueous electrolytes. These include the inherently poor solvation of generated protons, which is a characteristic limitation of aprotic solvents like DME due to their limited hydrogen bonding capabilities, often leading to sluggish proton transfer kinetics (46). Additionally, lower ionic conductivity compared to aqueous systems is a common feature of many nonaqueous electrolytes, resulting in increased ohmic drop (47). Furthermore, large interfacial resistances can arise from the formation of solid-electrolyte interphases or other surface passivation layers at the electrode–electrolyte interface, impeding charge transfer (48). The relatively limited ability of dimethoxyethane (DME) to stabilize highly charge-separated intermediates due to its lower dielectric constant can also contribute to sluggish charge transport kinetics at the electrode–electrolyte interface, leading to additional overpotentials (49). Moreover, gas crossover and possible flooding in mesh-based electrodes further exacerbate kinetic limitations.

Fig. 2.

(A–F) Tafel plots corresponding to: (A) 30 SS Mesh, (B) 350 nm Pt coated 30 SS Mesh, (C) 350 nm Ni coated 30 SS Mesh, (D) 80 SS Mesh, (E) 350 nm Pt coated 80 SS Mesh, (F) 350 nm Ni coated 80 SS Mesh. (Operating conditions: electrolyte – 0.5 M Ca(ClO₄)₂ + 0.065 M EtOH in DME; 10 sccm H₂ flow, Counterelectrode (cathode): 3 mm thick Ni foam and Working Electrode (anode): various tested anodes.) (G) NH₃ Faradaic efficiencies and partial current densities over four consecutive runs at −10 mA/cm², evaluating electrode stability (350 nm Pt coated 80 mesh) while changing the electrolyte between runs (Operating conditions: continuous commercial electrolyzer, electrolyte – 0.5 M Ca(ClO₄)₂ + 0.065 M EtOH in DME; 5 sccm N₂ and H₂ pressure, Working Electrode: Ni foam cathode and Counterelectrode: 350 Pt coated stainless steel mesh anode). (H) Voltage vs. time profiles for 0.5 mm thick Ni foam at a current density of −1 mA/cm² in continuous commercial electrolyzer. (I) Tafel plots for 0.5 mm thick Ni foam.

The Pt-coated 80 SS mesh and Ni coated 80 SS mesh offered the most promising results, demonstrating faster kinetics, effective backpressure maintenance, minimal leakage, and significantly improved voltage stability for at a current density of –1 mA cm⁻² (Fig. 1 F and G). Furthermore, the Pt-coated 80 SS mesh demonstrated longer operational stability and faster kinetics than the Ni-coated mesh and achieved an ammonia FE of 9.5%, underscoring the benefits of Pt surface modification. To further probe its performance under higher current densities, the reaction was carried out for 90 min using a 350 nm Pt-coated SS Mesh 3 at a total current density of –10 mA cm⁻². Under these conditions, we achieved a FE of 9.4 ± 1.5%. To test the stability of the electrodes, we conducted three additional consecutive runs, changing only the electrolyte between runs. As shown in Fig. 2G, the FE and CD remained nearly constant across all four runs, demonstrating the electrodes’ stability and reusability without disassembling. These electrodes offer a simple yet effective design and demonstrate long-term stability, significantly outperforming the 5 to 8 min of operation reported by Fu et al. (7) using simpler anodes. Additionally, they exhibit excellent reusability, maintaining selectivity even after four cycles. The ability to operate consistently across multiple runs highlights the robustness of the anode, making it well-suited for continuous operation while minimizing the need for frequent maintenance. Ammonia was quantified using NMR spectroscopy as shown in SI Appendix, Fig. S4.

Continuous-Flow PEEK Electrolyzer Setup.

The electrolyzer assembly presented multiple operational challenges, primarily related to material stability and gas–liquid interface. High gas flow rates of 75 sccm were necessary to prevent liquid seepage, making isotope labeling impossible and leading to inefficient nitrogen utilization. The graphite plates exhibited corrosion due to prolonged exposure to the organic electrolyte, which altered the serpentine gas flow channels. However, the PEEK block used for our electrolyte flow channel demonstrated significant chemical stability against DME and mechanical integrity. To overcome these challenges, we exclusively redesigned the flow cell using PEEK material. A schematic of the redesigned cell components is shown in SI Appendix, Fig. S2A, and the actual setup is shown in Fig. 3A. The cathodic chamber in the continuous PEEK electrolyzer was designed to securely hold the Ni foam cathode within a groove, ensuring an air-tight seal upon assembly and preventing electrolyte evaporation. This improved design also preserves the structural integrity of the Ni foam by avoiding compression, thereby maximizing its surface area and allowing unobstructed gas flow through the electrode. The electrolyzer’s electrical design included two key features to enhance performance: 1) power connections were placed directly next to the electrode surface to reduce losses, and 2) the cell body was constructed entirely from nonmetallic materials to ensure full electrical isolation. This design effectively eliminated any risk of short circuits between the current collectors and structural components while maintaining low-resistance contact with the catalyst layer.

Fig. 3.

(A) Continuous PEEK electrolyzer for continuous CaMAS. (B) NH₃ Faradaic efficiencies and NH₃ partial current densities at different applied current densities. (C) Counterpotential and working electrode potential at different applied current densities for benchmarking experiments. (Operating conditions: Continuous PEEK electrolyzer, electrolyte – 0.5 M Ca(ClO₄)₂ + 0.065 M EtOH in DME; 5 sccm N₂ and H₂ pressure, working electrode (cathode): 3 mm thick Ni foam and counterelectrode (anode): 0.5 mm thick Ni foam, Ag wire pseudo-reference electrode.) (D) NMR spectra of the solutions from the aforementioned experiments. The distinct proton-splitting patterns observed in the spectra from ¹⁵N₂ and Ar feed gas provide evidence that ammonia is generated from the reduction of the feed gas. (E) Ammonia concentrations in the electrolyte solutions were measured by NMR following experiments using ¹⁵N₂, ¹⁴N₂, and Ar as feed gases. −100 mA cm⁻² current was applied for 1 h in experiments with current, while in control experiments, no current was applied for the same duration. Error bars represent the SD in the quantification assay, based on n = 3 measurements.

Another challenge in the previous setup was the use of a 350 nm thick platinum coating on the electrode. While this configuration delivered good performance and allowed for electrode reuse, the high cost associated with platinum raised concerns regarding scalability and economic viability. As a result, efforts were directed toward exploring alternative HOR catalysts that do not rely on platinum-group precious metals. The development of nonnoble metal catalysts for the HOR is a major focus in the development of anion exchange membrane fuel cells. Among the candidates, nickel stands out due to its earth abundance and promising catalytic activity. Its suitability for HOR is largely attributed to its favorable hydrogen adsorption characteristics (50–52). Specifically, the strong interaction between the d-band of Ni and adsorbed hydrogen facilitates the subsequent steps in the HOR pathway. Additionally, nickel’s intrinsic oxophilicity enables it to modulate the hydroxyl binding energy of platinum when used in bimetallic systems, potentially enhancing the overall HOR performance (50). Additionally, Ni exhibits excellent chemical and mechanical stability in organic solvents such as DME, making it well-suited for use in systems where traditional noble metals may degrade or leach (50). These properties collectively contribute to Ni’s emerging role as a viable alternative to platinum in HOR applications.

As discussed previously the Tafel analysis showed that nickel and platinum exhibited good HOR activity on SS meshes, as indicated by their lower Tafel slope [350 nm Ni coated 30 SS Mesh ~ 0.32 V/dec, 350 nm Ni coated 80 SS Mesh ~ 0.211 V/dec]. However, the mesh structure introduced structural instabilities during operation. The gas–liquid interface was not consistently maintained, leading to periodic electrolyte flooding into the gas channel. Furthermore, the 350 nm coating thickness required a substantial amount of nickel/platinum, leading to significant consumption of the target. To address this issue, nickel foam was evaluated as an alternative anode material. Its interconnected porous structure provided better mechanical stability and a more robust gas–liquid interface, significantly reducing electrolyte crossover. Nickel foam offered comparable HOR activity without sputter coating, making it a more economical and scalable choice for anode design in this system. Tafel analysis was performed to rigorously evaluate the HOR kinetics on Ni foam for comparison with other anode materials. It exhibited stable operation and the lowest Tafel slope of 0.18 V/dec, indicating the most favorable HOR kinetics in our system (Fig. 2 H and I). For comparison, Pt-coated SS meshes and Pt-coated Ti mesh showed Tafel slopes of 0.199 V/dec (80 SS mesh), 0.26 V/dec (30 SS mesh), and 0.34 V/dec (Pt-coated Ti screen), while bare stainless steel meshes performed poorly, consistent with their known lack of catalytic activity and passivation behavior. Importantly, the superior performance of Ni foam is attributed not just to its catalytic nature but also to its high surface area and 3D porous architecture, which enhance gas diffusion and reduce interfacial resistance key factors in overcoming the mass transport limitations of nonaqueous electrolytes (50–52). Nickel is known to facilitate HOR via the Volmer–Heyrovsky mechanism in alkaline media (50–52), where hydrogen adsorption and subsequent electron transfer occur efficiently on its surface [1,2]. While a similar pathway may be plausible in nonaqueous electrolytes due to comparable proton-deficient conditions, differences in solvation, proton availability, and interfacial structure preclude direct mechanistic transfer. Further studies are needed to confirm the operative mechanism in organic systems. In contrast, stainless steel lacks intrinsic catalytic activity and suffers from oxide layer passivation, while platinum-coated SS meshes, although improved, still faced issues with nonuniform coating, flooding, and contact resistance at the interface. These results firmly establish nickel foam as the optimal anode material in our configuration, balancing catalytic activity, structural integrity, and electrochemical efficiency. Its low Tafel slope directly correlates with reduced overpotential requirements, enhanced system stability, and improved overall performance for ammonia electrosynthesis.

We tested 0.5 mm nickel foam as an anode in our electrolyzer setup. Nickel foam demonstrated better electrochemical cell stability and facilitated operation at lower voltages due to its high surface area, as it is a foam structure. It also effectively retained the electrolyte, preventing it from seeping into the gas flow channels while the back pressure was maintained. As corroborated in the literature, nickel foam has been widely studied as a gas diffusion electrode due to its unique structural and electrochemical properties, which enhance various reactions. Its high surface area provides abundant active sites, improving catalytic activity, while its excellent electrical conductivity ensures efficient charge transfer. The 3D porous structure allows for controlled gas diffusion and electrolyte retention, preventing flooding and maintaining back pressure—both critical for stable operation. Additionally, Ni foam is mechanically robust, making it durable under harsh electrochemical conditions. Its surface has also been functionalized with metal oxides or transition metals to enhance performance further in literature. Ni foam-immobilized Ni₆ nanoclusters have shown high current densities for the oxygen evolution reaction (53). Beyond this, nickel foam has been successfully employed in alkaline secondary batteries (54), direct borohydride fuel cells (55), and direct liquid fuel cells and electrolyzers (56), further supporting its effectiveness as a GDE material. Product analysis confirmed ammonia formation with enhanced selectivities, validating the suitability of nickel foam as an anode material for this system.

Before performing benchmarking experiments with the optimized anode, it was necessary to carefully evaluate the electrolyte composition. Calcium perchlorate is commercially available only in its hydrated form, and even after vacuum drying, the electrolyte retained approximately 3 ± 0.5 wt% water. This residual water can influence the reaction environment and potentially interfere with calcium-mediated ammonia synthesis. To evaluate the influence of the intrinsic ~3 wt% water content from the hydrated salt on ammonia synthesis, we systematically varied the water concentration in the electrolyte (SI Appendix, Fig. S34). When the electrolyte was further dried using a desiccant, reducing the water content to ~0.34 wt%, the FE increased to ~45%. In comparison, the standard electrolyte containing ~3 wt% water (dried under vacuum) resulted in ~35% FE. However, increasing the water content to ~9 wt% and ~14 wt% led to a decrease in FE to ~25 and ~22%, respectively (SI Appendix, Fig. S34). These results indicate that excess water adversely affects ammonia production, likely due to parasitic reactions between Ca metal and H₂O, forming Ca(OH)₂ rather than allowing Ca to form Ca₃N₂. This reduces the availability of active sites for nitrogen fixation. Therefore, although careful control of water content is essential, this approach was not entirely reliable, as water uptake varied depending on desiccant efficiency and exposure time, making it difficult to maintain a consistent water concentration across experiments. To ensure reproducibility, we used vacuum-dried salts with a constant ~3% water content for all subsequent experiments. Moving forward, the development or use of fully dehydrated calcium salts capable of enabling efficient Ca plating and nitrogen reduction should be considered a key direction for further optimization in calcium-mediated ammonia synthesis.

Additional experiments were conducted using the optimized flow cell. In these experiments, 3 mm thick nickel foam was used as the cathode, while 0.5 mm thick nickel foam served as the anode. Benchmarking was performed using the continuous-flow PEEK electrolyzer, as shown in Fig. 3A, across a current density range of −25 to −150 mA/cm². As anticipated, the nitride formation rate increased with current density, aligning with previous studies that have reported enhanced nitrogen fixation in lithium-ion batteries at higher plating current densities (15, 57). At approximately −100 mA/cm², the FE for ammonia synthesis reached its peak, i.e., 34.5 ± 2%. Beyond this current density, the FE began to decline as seen in Fig. 3B. This decrease can be attributed to nitrogen transport limitations, which likely hinder the availability of reactive nitrogen species at elevated current densities (15). Additionally, the decline in FE may result from excessive calcium plating at higher currents, where newly deposited layers accumulate on top of previous ones, rendering the underlying calcium inactive and reducing overall efficiency (15). Another possible explanation for the optimal current density observed is a localized depletion of dissolved nitrogen at the electrode surface due to mass transport limitations. In this scenario, the partial current density for ammonia synthesis approaches the mass transfer-limited regime, restricting further improvement in FE beyond this point. At higher current densities, the working electrode potential increases rapidly, which may suppress selective ammonia formation and promote side reactions such as solvent decomposition or hydrogen evolution, thereby lowering the FE. The variation of working and counterelectrode voltages with increasing current density is illustrated in Fig. 3C (15).

To assess the system’s stability in terms of operating potential and overall performance, long-term continuous experiments were conducted over a period of 56 h. The setup operated under chronopotentiometric conditions at a constant current density of −5 mA/cm². Throughout the experiment, the cell maintained a stable operating potential, indicating the robustness and durability of the system, as illustrated in Fig. 4A. Additionally, the ammonia FE and CD remained relatively consistent, around ~20% throughout the entire duration, as shown in Fig. 4B. Samples were collected every 4 to 5 h to monitor ammonia production across different time intervals and the ammonia concentration appears to increase steadily over the duration of the experiment, as shown in SI Appendix, Fig. S16E. These results confirm that the optimized setup is capable of sustaining ammonia synthesis over extended periods. To further assess the system’s performance under more stringent operating conditions, we conducted experiments at an elevated total cell current of −750 mA/cm². Impressively, the system maintained stable operation under these demanding conditions, achieving an ammonia partial current density of approximately −219 mA/cm² at a cell potential of ~28 V. The FE reached 29.2%, a highly promising value given the operational scale as shown in SI Appendix, Fig. S3. Electrochemical impedance spectroscopy also showed that the resistance remained essentially unchanged, with values of 10 ± 1 Ω before electrolysis and 9 ± 1 Ω after 1 h of operation. This minimal variation indicates that the system remains stable under continuous operation also shown by the long-term study. These results not only demonstrate the system’s long-term stability but also confirm its ability to sustain high-throughput operation, reinforcing its potential for practical industrial implementation.

Fig. 4.

(A–E) 56 h long-term experiment: (A) Working electrode potential (V vs. Ag wire) for the long-term run, (B) NH₃ Faradaic efficiencies at different time intervals. (C) High-resolution XPS scan of the postelectrolysis electrode confirming the presence of N (1s). (D) High-resolution XPS scan of the postelectrolysis electrode confirming the presence of Ca⁰ (2p). (E) NMR spectra of the post electrolyte after 56 h 9 min of operation from long-term experiments. (Operating conditions: Continuous PEEK electrolyzer, electrolyte – 0.5 M Ca(ClO₄)₂ + 0.065 M EtOH in DME; 5 sccm N₂ and H₂ pressure, 3 mm thick Ni foam cathode and 0.5 mm thick Ni foam anode, −5 mA/cm².) (F–K) Characterization of electrode (F) EDS spectrum of postelectrolysis electrode showing the presence of Ca and N. (G) High-resolution XPS scan of the postelectrolysis electrode confirming the presence of Ca⁰ (2p). (H) High-resolution XPS scan of the postelectrolysis electrode confirming the presence of N (1s). (I) Raman microscopic image of in situ nickel foam during the reaction. (J) Raman spectra of the Ni foam electrode during the reaction. (K) In situ FTIR spectra showing the symmetric bending mode of ammonia. Reaction Conditions: 0.5 M Ca(ClO₄)₂ in DME + 0.065 M EtOH, initially saturated with N₂, followed by continuous 5 sccm N₂ flow, applied current: −50 mA/cm².

Quantitative analysis of ammonia was conducted using ¹H NMR spectroscopy, following a previously published protocol that ensures the measured NH₃ arises from electrochemical N₂ reduction and not from air or other potential sources (8). As shown in SI Appendix, Fig. S4, the triplet peak associated with ammonia containing the most abundant ¹⁴N₂ isotope (spin 1 nucleus) appears at 6.83 ppm with a coupling constant of 45 Hz. ¹⁵N₂ isotope labeling and rigorous control experiments were performed to ensure that NH₃ is produced by the electrochemical reduction of N₂ and not from other sources. The experiment was conducted with isotope labeled nitrogen (¹⁵N₂) at a current density of −100 mA/cm²; this condition resulted in the highest ammonia FE. The isotope-labeled samples were quantified using ¹H NMR, and the spectra are shown in Fig. 3D. These experiments yielded a FE of 34 ± 3%, with an ammonia current density of −34 ± 3 mA/cm² at −100 mA/cm². The NMR spectrum of ammonia generated from ¹⁵N₂ displays a characteristic doublet at 6.78 ppm, with a coupling constant of 100 Hz. The coupling constant observed here for ammonia differs from the values reported by Anderson et al. Coupling constants such as are not purely intrinsic properties of the isolated molecule. Instead, they can be subtly influenced by several environmental factors. These include the presence of explicit hydrogen bonds, which modify the electronic distribution; solvent effects that arise from differences between gas-phase and liquid-phase environments; and electronic polarization, where the surrounding medium distorts the electronic cloud of the solute. Furthermore, geometry relaxation, small adjustments in bond lengths and angles as the molecule adapts to its local environment can also lead to measurable changes in coupling constants. Collectively, these effects demonstrate that solvent properties, hydrogen bonding, and related local structural factors can indeed affect the experimentally observed coupling constants (58). Open circuit control experiments were conducted using ¹⁴N₂ and ¹⁵N₂ where no potential was applied, and electrolyzer was set up using N₂ and H₂ flow. ¹H NMR spectra of both pre-electrolyte and postelectrolyte samples showed no ammonia, as shown in Fig. 3D. Another control experiment used argon (Ar) gas flow in the reactor instead of N2 at the same operating conditions. Postelectrolyte analysis using ¹H NMR showed no ammonia peaks for Ar experiment, as shown in Fig. 3 D and E.

Cyclic voltammetry (CV) was also performed using the same reaction setup to examine the electrochemical behavior of calcium in a 0.5 M Ca(ClO₄)₂ electrolyte dissolved in DME, with 0.065 M EtOH serving as a proton donor. The CV scan, conducted within a potential range of −2.5 V to 2.5 V vs. Ag wire at a scan rate of 5 mV/s, reveals distinct cathodic and anodic peaks associated with the deposition and stripping of Ca, respectively, as illustrated in SI Appendix, Fig. S5. During the cathodic sweep (59), a notable current onset at approximately −1.63 V vs. Ag indicates the reduction of Ca²⁺ ions to metallic Ca on the working electrode. In the reverse anodic sweep, an oxidation peak appears around 1.54 V vs. Ag, corresponding to the dissolution of the deposited calcium back into the electrolyte as Ca²⁺. To confirm the occurrence of the HOR at the anode, linear sweep voltammetry (LSV) was performed under two conditions: with argon (Ar) at the anode and with hydrogen (H₂) at the anode. The LSV profile with H₂ showed a significantly steeper current response compared to the profile with Ar, confirming the HOR’s activity at the anode as shown in SI Appendix, Fig. S5. This steep response with H₂ indicates the efficient oxidation of hydrogen, thereby verifying its role in supplying protons for the electrochemical reaction (12, 15).

The postreaction characterization of the electrode was conducted using scanning electron microscopy coupled with energy dispersive spectroscopy (SEM-EDS) and X-ray photoelectron spectroscopy (XPS) to gain insights into the electrode surface morphology and composition, respectively. SI Appendix, Fig. S15A shows the SEM-EDS elemental mapping results, which demonstrate a strong overlap between the distributions of calcium and nitrogen, suggesting the possible formation of calcium nitride on the electrode surface. In contrast, the mappings of nickel (Ni), chlorine (Cl), and oxygen (O) exhibit overlapping regions that are spatially distinct from the Ca and N distributions, indicating that these elements are associated with different surface phases or compounds. The survey scan also confirms the presence of Ca and N as shown in Fig. 4F. Furthermore, an XPS analysis was performed to see the presence of metallic calcium and nitrogen. As illustrated in Fig. 4G, the XPS analysis confirms the presence of calcium in two oxidation states: metallic calcium (Ca⁰) and calcium ions (Ca²⁺). Additionally, the XPS spectra reveal the presence of nitrogen, as depicted in Fig. 4H. These findings collectively suggest the formation of a calcium nitride layer on the electrode surface, which could play a crucial role in the overall electrochemical process. To further verify that the observed nitrogen signal originates from calcium nitride, XPS was performed on commercially available Ca₃N₂ powder as a reference. The N 1s spectra of both the reference Ca₃N₂ and the postreaction electrode exhibit overlapping peaks centered at approximately 400.3 eV, indicating a comparable chemical environment and supporting the presence of calcium nitride in the postreaction sample as shown in SI Appendix, Fig. S19 A and B (60). Ca 2p peaks were also analyzed to precisely confirm the presence of Ca₃N₂. As shown in SI Appendix, Fig. S19E, the postreaction electrode exhibits peaks at 347.5 eV (Ca 2p₃/₂) and 351.0 eV (Ca 2p₁/₂), which align well with those of pure calcium nitride. However, the peaks are slightly broadened, suggesting the presence of a small amount of metallic Ca⁰. Furthermore, Auger electron spectroscopy was conducted to examine the Ca Auger transitions in both the reference and postreaction samples. The Ca Auger peaks observed near 290 eV in both cases show strong agreement, providing additional confirmation of the formation and retention of calcium nitride on the electrode surface following the reaction as shown in SI Appendix, Fig. S19 C and D. Pristine Ni foam was also characterized as a control. As shown in SI Appendix, Fig. S27, it displays the expected Ni 2p and O 1s peaks. To assess whether any changes arise from simple electrolyte exposure, the pristine Ni foam was soaked in the electrolyte for 1 h—the same duration as the reaction—and subsequently analyzed. As shown in SI Appendix, Fig. S28, the spectrum shows Ca²⁺ features corresponding to the Ca 2p₁/₂ and Ca 2p₃/₂ peaks, but the narrow Ca 2p₃/₂ peak indicates that metallic Ca⁰ is absent.

In order to investigate the structural and compositional changes to the anode following long-term electrochemical operation, the Ni foam anode was analyzed after 56 h of continuous electrolysis at –5 mA cm⁻². SEM imaging revealed that the Ni foam retained its overall structural integrity, indicating good mechanical stability under prolonged operation. However, in the case of higher current density experiments (e.g., –750 mA cm⁻²), minor physical damage such as localized pitting and surface roughening was observed, likely due to increased gas evolution and stress at the electrode surface. Energy-dispersive X-ray spectroscopy (EDS) was conducted with SEM to evaluate the elemental composition of the anode surface. Elemental mapping showed a relatively uniform distribution of calcium and oxygen, consistent with the potential formation of calcium-based compounds or byproducts. Despite these deposits, no significant delamination or large-scale degradation was evident, demonstrating the electrode’s robustness. Together, these findings highlight the mechanical durability and chemical resilience of Ni foam anodes during extended operation, even under high current stress, while also providing insights into surface modifications induced by electrolysis.

Postreaction characterization of the electrode used in the long-term electrolysis experiment (56 h at −5 mA cm⁻²) was conducted using SEM-EDS and XPS to assess its structural and chemical stability. As shown in Fig. 4 C and D, the XPS analysis revealed the presence of a distinct N 1s signal and metallic calcium (Ca⁰), indicating the formation of calcium nitride on the electrode surface. SEM images confirmed the structural integrity of the electrode, while EDS elemental mapping showed overlapping distributions of calcium and nitrogen, further supporting the presence of calcium nitride as shown in SI Appendix, Fig. S16G. Additionally, the electrode used in the high current electrolysis experiment (–750 mA cm⁻²) was characterized to evaluate its structural robustness, as shown in SI Appendix, Fig. S3. SEM imaging revealed the Ni foam morphology postelectrolysis and EDS elemental mapping confirmed the presence of nitrogen on the electrode surface. Minor surface irregularities were observed, likely due to the high applied potential; nevertheless, the electrode retained its overall structural integrity.

The SEI is a highly complex structure, making its characterization challenging. In situ studies offer a valuable approach to understanding its formation and composition, although their interpretation remains challenging. To investigate the SEI in our system, we performed in-situ Raman spectroscopy to probe the possible formation of calcium nitride during calcium plating. A specialized electrochemical cell was used for the setup, with a platinum counterelectrode and a 3 mm thick nickel foam working electrode. Nitrogen gas was continuously bubbled using a syringe to minimize disturbances and prevent liquid sloshing, as seen in Fig. 4I. Raman spectra were recorded for 30 min of reaction at different time intervals. A distinct peak corresponding to calcium nitride is observed at approximately 260 and 290 cm⁻¹, as shown in Fig. 4J, consistent with the findings of Heyns et al. (61) In-situ Raman spectroscopy analysis confirms the successful plating of calcium and the subsequent formation of calcium nitride on the Ni foam cathode.

Furthermore, to directly probe the formation of NH₃ species during the reaction, we performed in-situ FTIR analysis. Spectra were collected at various time intervals over the course of the experiment. As noted above, two distinct spectral features were consistently observed at 933 and 975 cm⁻¹, which were assigned to the symmetric bending mode vibrations of NH₃, providing clear spectroscopic evidence for ammonia generation under the studied conditions (Fig. 4K). Due to the relatively high concentrations of solvent and reactants in our system, it was challenging to resolve additional FTIR peaks corresponding to possible species within the SEI. Nonetheless, the observed ammonia peaks support the proposed reaction mechanism and complement the Raman and electrochemical analyses presented in this work (62).

To assess changes in the electrolyte after electrolysis, we examined the system using ¹H NMR and in situ FTIR spectroscopy (SI Appendix, Fig. S18). New peaks in the postelectrolysis NMR suggest formation of small organic species such as methyl formate, vinyl compounds, and aldehydes, likely from partial oxidation or decomposition of DME and ethanol. The FTIR spectra further reveals new C-H stretches characteristic of aldehyde groups. These observations confirm that controlled electrolyte degradation occurs, contributing to SEI formation—a known phenomenon in metal-mediated ammonia synthesis and battery systems (43, 63, 64). Importantly, we observed that DME degrades less severely than other solvents like THF (10), which produce more extensive polymeric byproducts. This partial degradation appears necessary to stabilize the electrode interface while still allowing sustained operation.

Conclusion

This work advances the development of a continuous calcium-mediated approach for electrochemical ammonia synthesis and highlights key insights and engineering challenges in the Ca-NRR. The significant improvement in ammonia FE from 9.4 ± 1.5% in the continuous commercial electrolyzer to over 34.5 ± 2% in the continuous PEEK electrolyzer setup underscores the importance of reactor design and electrode configuration in optimizing performance. Using a cost-effective nickel-based electrode further demonstrates the potential of scaling up Ca-mediated ammonia synthesis without relying on expensive materials. Catalyst characterization and isotope labeling were conducted to validate N₂ activation further. In-situ Raman spectroscopy was utilized to monitor the formation of calcium nitride on the system during the reaction. The system demonstrated excellent performance over 56 h of continuous electrolysis, maintaining a stable cell potential and an average ammonia selectivity of approximately 20%. High current density experiments further highlighted the system’s robustness, achieving an ammonia partial current density of ~219 mA cm⁻² with a FE of ~29%, underscoring its potential for scalable and energy-efficient ammonia production. Further investigation is required to elucidate the competition between EtOH reduction and N₂ activation, as well as the role of the Ca-rich layer in facilitating nitrogen fixation. Understanding the composition, structure, and stability of calcium nitride and related species will be critical for enhancing selectivity and efficiency. Additionally, insights from SEI engineering in battery research may aid in improving the stability and ionic conductivity of the Ca SEI layer. This study motivates the exploration of other metal alternatives such as magnesium, barium, strontium, and sodium for nitrogen reduction, potentially expanding the scope of efficient and scalable electrochemical fertilizer production technologies. This approach also paves the way for truly sustainable ammonia synthesis by utilizing renewable electricity to power the reaction and relying on abundant resources like air and water. This not only enhances the feasibility of green ammonia production but also holds great potential for enabling decentralized and environmentally friendly manufacturing.

Materials and Methods

Electrochemical Experiments.

Continuous electrolyzer setup.

Ammonia synthesis was conducted in a three-compartment, continuous-flow cell using PEEK housing and nickel foam electrodes. Calcium ions were reduced electrochemically to form Ca, which reacted with N₂ to generate Ca₃N₂. Ethanol served as a proton donor, and H₂ oxidation at the Pt anode regenerated protons. Chronopotentiometry was performed at 100 mA/cm² for 1.5 h using a Biologic SP-300 potentiostat.

PEEK reactor optimization.

A fully PEEK-based cell was designed with serpentine gas channels, integrated sealing features, and a three-electrode configuration (Ag wire pseudo-reference). Electrolyte flow was maintained at 1 mL/min, and gas flows at 5 sccm. iR compensation (85%) was applied. Long-term (−5 mA/cm², 56 h) and high-current (750 mA) runs were performed to assess durability and productivity.

Tafel experiments (HOR).

To evaluate hydrogen oxidation activity, LSVs were performed in an H-cell with H₂ bubbling and various anode materials. Potentials were referenced to Ag wire and internally calibrated using the Fc/Fc⁺ redox couple. Overpotentials were used to extract Tafel slopes for each material.

Ammonia quantification.

NH₃ concentration was determined using the Indophenol blue method with UV-vis spectroscopy at 632 nm. A standard additions method was used to correct for matrix effects, and absorbance–concentration correlations were used to determine NH₃ yield and FE.

NMR analysis.

Quantitative NMR was used to distinguish between ¹⁴NH₃ and ¹⁵NH₃. Postelectrolysis solutions were acidified, mixed with DMSO-d₆, and analyzed using a 500 MHz NMR with water suppression. Spectral integration was used to quantify NH₃ concentration.

Electrode characterization.

Postelectrolysis electrodes were analyzed using SEM-EDS, XPS, and in-situ Raman. SEM-EDS provided spatial elemental distribution, while XPS identified Ca and N chemical states. Raman spectroscopy tracked Ca₃N₂ formation in real-time under operating conditions.

Detailed experimental procedures and characterization data are provided in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.