Magnetic field–enhanced alkaline water electrolysis from laboratory to industry

Significance

Industrial alkaline water electrolysis (AWE) suffers from the progressive depletion of dynamic active sites and rapid dissipation of local active ion concentration, limiting efficiency, and scalability. Herein, we propose a magnetic field–enhanced AWE (ME-AWE) strategy that induces Lorentz-force-driven convection to enrich iron ions locally at the anode, overcoming intrinsic concentration limitations without increasing bulk impurities. The mechanism is quantitatively validated and interpreted by controlled electrochemical experiments and multiphysics simulations, and further implemented in an industry-scale electrolyzer with integrated magnetic field design. The ME-AWE achieves a 24.9% enhancement in hydrogen production at 1.8 V and 80 °C compared to the conventional configuration. Techno-economic analysis indicates a 10.6% reduction in AWE capital cost in off-grid wind-to-hydrogen applications.

Abstract

The fundamental limitation of water oxidation in alkaline water electrolysis (AWE) lies in the progressive depletion of dynamic active sites and rapid dissipation of local active ion concentration, a critical thermodynamic constraint that conventional electrochemical strategies struggle to overcome. Here, we propose a magnetic field–enhanced AWE (ME-AWE) strategy that enables in-situ directional enrichment of local active iron ions near the anode, thus breaking the concentration limit and boosting the coverage of dynamic iron active sites without elevating bulk impurity levels. Laboratory-scale three-electrode experiments confirm improved oxygen evolution reaction (OER) kinetics by the magnetically induced optimization of iron incorporation dynamics and the increase in the coverage of dynamic iron active sites. A multiphysics model coupling magnetic field, fluid flow, mass transport, and electrochemical reaction is developed to spatially interpret the mechanism, revealing that the magnetic modulation enables OER kinetics at 0.3 ppm iron equivalent to those at tenfold higher concentration without a magnetic field. Guided by mechanistic insights, an industry-scale ME-AWE device is designed and implemented, achieving a 24.9% increase in hydrogen output at 1.8 V. Techno-economic analysis demonstrates that the ME-AWE strategy enables a 10.6% reduction in total AWE capital cost for large-scale hydrogen production plants using off-grid wind power. Built with existing industrial infrastructure, this ME-AWE strategy offers a scalable and low-cost solution for improving AWE efficiency and advancing impurity-assisted catalysis in green hydrogen production.

Hydrogen is increasingly seen as a key enabler in the global transition to a climate-neutral economy, especially for the decarbonization of hard-to-abate sectors (1, 2). Coupling water electrolysis with renewable electricity offers a promising route for producing green hydrogen while stabilizing intermittent energy supply (3, 4). According to the International Energy Agency (IEA), the global electrolyzer capacity reached 1.4 GW by 2023 and could surge to over 230 GW by 2030 (5), reflecting the rapid scale-up of green hydrogen deployment (6). Within this momentum, alkaline water electrolysis (AWE) remains the most widely adopted electrolysis technology, owing to the technological maturity and cost-effectiveness (7). However, as AWE scales to industrial applications, the energy conversion efficiency emerges as a major performance limiting factor under industrial conditions (8, 9).

Fundamentally, water electrolysis constitutes a tightly coupled multiphysics process encompassing electrochemical reactions, momentum transport, mass transport, and heat transfer (10, 11), with the dynamic interplay of these physical fields collectively governing the performance of AWE devices (12). In terms of the internal field regulation, efforts at the material level focus on modulating electrochemical processes by tuning electrode physicochemical properties (13, 14). For the electrolyzer scale, flow field optimization in bipolar plate channels based on the flow field model, demonstrates measurable improvements in bubble management and hydrogen production rates (15, 16). Considering the multienergy coordination of the AWE device, control strategies for optimizing operational parameters under varying conditions have been proposed to ensure the efficient performance (17, 18). Despite efforts on internal field regulation, the potential for further enhancement remains constrained by limited degrees of physical field tunability.

External field regulations [for example, magnetic field (19), ultrasonic field, (20) and super gravity field (21)], as flexible and contactless approaches, offer additional degrees of control to enhance electrochemical performance (22–24). Among these, magnetic field is particularly promising due to its simple implementation and multiple enhancement mechanisms (25, 26), including but not limited to the magnetohydrodynamic (MHD) effect (27), Kelvin force (28), magnetic hyperthermia, (29) and spin polarization (25). In AWE devices, the mass transport of redeposition-capable metal species that maintain catalytic sites plays a decisive role in system performance, in conjunction with reactants and gaseous products (30, 31). Although recent visual and quantitative research has shown improved ion transport and reaction performance in some mass-transport-limited reactions, the improvement in OER [rate-determining step in AWE (13)] remains marginal (32). Currently, studies on magnetic enhancement are largely confined to laboratory-scale investigations and overlook industrial constraints, resulting in the absence of research on magnetic regulation of electrolyte impurities (31). In industrial AWE devices, iron ions (the dominant cationic impurities) are increasingly recognized as critical performance modulators by dynamically forming and regenerating iron active sites on anode catalysts (33, 34). However, constrained by the thermodynamics, iron active sites at the anode undergo inevitable progressive depletion, posing a persistent challenge for existing electrochemical remediation strategies (35). From a kinetic perspective, elevating iron ion concentration enhances AWE performance (36), but at the expense of accelerated diaphragm and BoP component corrosion (37). While localized enrichment of iron ions at the anode surface is desired, conventional AWE devices inevitably increase bulk concentration, thereby compromising durability and elevating lifecycle costs (38).

To address these limitations, we propose a magnetic field–enhanced AWE (ME-AWE) strategy that enables spatially controlled enrichment of dissolved iron ions near the anode. In contrast to the previous magnetic field–assisted OER studies that primarily focused on MHD-driven bubble dynamics (27, 39), interfacial structural modulation (40, 41) and electronic spin polarization (25, 42), this work reveals a magnetically driven local enrichment of trace iron ions in the electrolyte, enhancing the coverage of dynamic iron active sites and thus accelerating the OER kinetics. Electrochemical tests and characterizations consistently confirm the magnetically enhanced OER kinetics by mediating the dynamics of iron incorporation, which boosts the coverage of dynamic iron active sites and achieves a 44 mV advantage at 300 mA/cm². Multiphysics simulations further reveal that the magnetic modulation enables OER kinetics at 0.3 ppm iron equivalent to those at tenfold higher concentration without a magnetic field. Applied in an industrial AWE device, the ME-AWE delivered a 24.9% enhancement in hydrogen production at 1.8 V and 80 °C compared to the conventional nonmagnetic AWE (NM-AWE) configuration. To evaluate the economic potential of large-scale ME-AWE deployment in green hydrogen scenarios, a techno-economic assessment is conducted, further demonstrating its superior profitability and practical feasibility when coupled with fluctuating renewable power sources. The proposed ME-AWE offers a noncontact and controllable strategy, providing valuable insights for advancing impurity-assisted catalysis and enhancing AWE efficiency. Importantly, compatible with existing industrial processes, it holds strong potential for translation from laboratory investigation to industrial deployment in green hydrogen production.

Results

Proposed Mechanism of ME-AWE.

In industrial AWE devices, the electrolyte is in direct contact with the anode surface, serving as the medium that delivers dissolved iron ions to the OER sites. According to the theory of dynamic iron active sites, the dissolved iron ions can dynamically deposit and dissolve in situ on the nickel-based anode surface during the OER process (35). Since OER kinetics highly depend on the availability of dynamic iron active sites, the local concentration of iron ions near the anode surface critically affects the overall reaction performance of the AWE devices.

Fig. 1 illustrates the behavior of dissolved iron ions and the modulation mechanism introduced by the external magnetic field. Under conventional NM-AWE condition (Fig. 1A), the concentration of iron ions in the electrolyte is extremely low and uniformly dispersed. Furthermore, due to the electric field effect, iron ions tend to drift away from the anode, leading to insufficient local concentration and limited iron deposition. Thereby, the beneficial effect of iron ions in promoting OER activity is not fully realized.

Fig. 1.

Mechanism of ME-AWE. (A) During OER, dissolved iron ions drift away from the anode under influence of the electric field, leading to minimal deposited iron on the surface. (B) Under the magnetic field, the Lorentz-driven stirring flow facilitates the enrichment of iron ions near the anode, thereby increasing the iron deposition on the anode surface.

Under ME-AWE condition (Fig. 1B), the disk-shaped unit structure commonly used in industrial electrolyzers enables the practical introduction of a magnetic field perpendicular to the electrode surface. The distortion of current lines at the electrode edges generates circumferential Lorentz forces on the electrolyte, as dictated by the left-hand rule. Notably, the current line distortion can be also induced by surface irregularities or insulating gas bubbles on the electrode surface, commonly resulting in the presence of circumferential Lorentz force. Accordingly, the strong stirring flow is induced, effectively confining iron ions to the vicinity of the anode. Furthermore, the circular flow generates centrifugal effects that create localized low-pressure regions at the anode center, thereby driving electrolyte flow vertically toward the electrode surface. The resulting flow pattern is particularly advantageous, enabling direct delivery of fresh dissolved iron from the bulk solution. Therefore, despite the drift under the electric field, the combined effect of circular and vertical flows substantially increases the iron ion concentration at the anode surface. This localized enrichment significantly boosts the coverage of dynamic iron active sites, enhancing the OER process. Notably, the ME-AWE can provide an effective strategy to increase the local concentration of iron ions at the anode without violating industrial constraints that limit the iron content in the electrolyte, meeting the long-term operational requirements of industrial AWE devices.

Laboratory-Scale Experimental Validation.

To quantitatively validate the ME-AWE, laboratory-scale three-electrode experiments were conducted, incorporating four experimental groups for systematic comparison (SI Appendix, Fig. S1). The comparison between iron-containing groups was employed to evaluate the magnetic effect by iron ion modulation. To clarify the mechanism of proposed magnetic effect, groups without added iron ions were designed as a controlled pair to assess the potential contribution of other magnetic field-induced mechanisms, such as effects on bubble dynamics and reactant mass transport kinetics.

Under industrial 30 wt% KOH electrolyte without added iron ions (dashed lines in Fig. 2A), applying a 0.18 T magnetic field has negligible impact on the electrocatalytic performance of OER, with only a 3 mV reduction in potential versus the reversible hydrogen electrode (vs. RHE) at 300 mA/cm². This result indicates the minimal other magnetic effects on OER under controlled conditions, consistent with the previous findings (32). In the presence of 3 ppm iron ions (solid lines in Fig. 2A), the polarization curves shift left markedly, indicating a strong catalytic enhancement of the nickel anode. Under this condition, the magnetic field group exhibits a pronounced 44 mV reduction in potential at the current density of 300 mA/cm², clearly demonstrating the effectiveness of iron-related magnetic field effect in enhancing OER performance. The electrochemically active surface area (ECSA) measurements exhibit comparable surface roughness of both groups containing 3 ppm iron ions (SI Appendix, Fig. S2) and the ECSA-normalized current density follows the same trend as that obtained based on the geometric area, indicating the enhanced intrinsic activity under magnetic field (SI Appendix, Fig. S3).

Fig. 2.

Validation of magnetic enhancement through electrochemical measurements and characterizations. (A) OER polarization curves in 30 wt% KOH electrolyte (IR compensated), with and without the addition of 3 ppm dissolved iron ions. (B) OER potential (IR compensated) at 400 mA/cm² during CV process. (C) Tafel slope comparison. (D) Nyquist plot comparison. (E) Contour plots of current density in the potential range of 1.20 to 1.35 V during the forward CV scan. (F and G) High-resolution XPS spectra of (F) Fe 2p and (G) Ni 2p. (H and I) From left to right: TEM image, high-angle annular dark-field (HAADF) image and elemental mapping (Ni-Fe overlap, Fe, Ni) of the anodes under (H) magnetic field and (I) nonmagnetic field conditions. (J and K) EPMA-WDS mapping of Fe and Ni for the anode under (J) magnetic field and (K) nonmagnetic field conditions. Note: (B–K) correspond to anodes tested and characterized under 3 ppm iron ion conditions.

To gain deeper insights into the catalytic enhancement induced by iron-related magnetic field effect, detailed investigations are conducted on iron-containing groups. During the cyclic voltammetry (CV) pretreatment, the trend of OER potentials reflect progressive anode activation and stabilization upon the addition of iron ions (Fig. 2B and SI Appendix, Fig. S4). The potential advantage gradually increases over the first 25 CV cycles and eventually stabilizes at approximately 44 mV. Tafel slope is determined by using the logarithm of current density against the potentials to evaluate the reaction kinetic improvement. The calculated Tafel slope under magnetic field is about 49.9 mV/dec, which is smaller than that without magnetic field application, indicating a superior electron transfer efficiency for OER (Fig. 2C). The electrochemical impedance spectroscopy (EIS) analysis results also exhibit a similar trend, with the charge transfer resistance (R_ct) and the oxygen intermediate species adsorption resistance (R_p) both decreasing by 10-30% across various applied potentials (Fig. 2D and SI Appendix, Fig. S5).

Compared to the nonmagnetic case, the magnetic field–assisted anode exhibits a significantly enhanced overall intensity in its oxidation peaks during the forward CV scan (Fig. 2E and SI Appendix, Fig. S4), indicating an accelerated formation and accumulation of NiOOH and consequently enhanced intrinsic activity. A pronounced double-peak feature emerges distinctly under magnetic field, which can be assigned to the oxidation of α- and β-Ni(OH)2 to γ- and β-NiOOH, respectively (43). Under magnetic field operation, the α-Ni(OH)₂/γ-NiOOH peak becomes significantly more intense than in the nonmagnetic case, indicating extensive interlayer-anion intercalation driven by the local charge imbalance arising from the promoted iron ion incorporation. During cycles 2-15, the α-Ni(OH)₂/γ-NiOOH oxidation peak progressively weakens, which suggests that the initially abundant and highly uneven iron incorporation becomes progressively more uniform under the repeated anodic-dissolution and cathodic-redeposition steps of CV cycling. In the subsequent cycles 16-40, the continued magnetically induced enrichment enables increasing iron uniform incorporation, leading to a gradual intensification of the α-Ni(OH)₂/γ-NiOOH oxidation peak and a progressive positive shift of the oxidation potential.

To further clarify the role of Fe and its interaction with Ni, XPS analysis after the 1st and 40th CV cycles under magnetic and nonmagnetic conditions was performed. The Fe 2p spectrum can be respectively fitted into a pair of peaks with Fe³⁺ 2p_3/2 and 2p_1/2 (Fig. 2F). After both 1st and 40th CV cycles, the Fe 2p peak under an applied magnetic field shifts toward higher binding energy compared to the nonmagnetic field case. This shift indicates not only the oxidation of Fe sites to higher valence states but also an enhancement in the Fe-O covalency within the Fe-O-Ni unit formed following Fe incorporation, thus strengthening the Fe sites (43). Consistently, the increased Ni³⁺/Ni²⁺ area ratio (Fig. 2G and SI Appendix, Table S1) suggests that neighboring Ni atoms are more readily driven into higher-valence, catalytically active states within this electronically coupled Fe-O-Ni unit. In addition, the Fe 2p shift of binding energy after 40th CV cycle (0.42 eV) is larger than that after 1st CV cycle (0.23 eV), indicating magnetic effect continues to promote progressive Fe incorporation during the CV process, consistent with the increasing potential advantage observed in the CV process (Fig. 2B).

Transmission electron microscope (TEM) and elemental mapping further reveal a more homogeneous morphology and uniformly distributed iron sites under the magnetic field condition (Fig. 2H). In contrast, without the magnetic field, the extremely low iron ion concentration near the anode surface tends to limit iron incorporation, leaving iron sites sparsely distributed, which seem mainly localized at edge sites (Fig. 2I). Furthermore, high-precision quantitative analysis using electron probe microanalysis (EPMA) equipped with a wavelength-dispersive spectrometer (WDS) reveals the mass-fraction distribution of Ni and Fe (Fig. 2 J and K). Under magnetic field operation, the average Fe content exhibits a 2.5-fold increase relative to the nonmagnetic condition, further corroborating the proposed magnetically induced mechanism.

Collectively, these results demonstrate that the magnetic field promotes local iron ion enrichment at the anode surface, thereby increasing the coverage of dynamic iron active sites and enhancing the intrinsic OER activity of commercial nickel foam. To further assess the generality of this mechanism, comparative tests were conducted on intrinsically active NiFe anodes. The chronoamperometry (CA) results show that the magnetic field significantly suppresses current-density decay (SI Appendix, Fig. S6), indicating the mitigation of dynamic iron net dissolution (44) and improved stability of iron-rich benchmark catalysts by magnetic modulation. Extending this evaluation, we replaced iron ions with an equivalent concentration of cobalt ions and observed reduced potential under magnetic field in the CV curves (SI Appendix, Fig. S7). Consistent with the increased cobalt content quantified by EPMA (SI Appendix, Figs. S8 and S9), these results further confirm that the magnetic field likewise promotes local cobalt ion enrichment and incorporation (45).

Multiphysics Mechanism Elucidation and Design Guidance.

Given the quantitatively validated magnetic mechanism, it is essential to establish a model that elucidates the internal coupled processes in depth, enables accurate performance prediction and provides guidance for magnetic field design. In particular, the water electrolysis involves multiphysics processes, whose coupling fundamentally determines the reaction performance. Therefore, a multiphysics model coupling magnetic field, fluid flow, mass transport, and electrochemical reaction was developed (SI Appendix, Fig. S10). Fig. 3A illustrates the three-dimensional model under the uniform magnetic field directed along the negative y-axis. The blue y–z plane is selected as the view slice for multiphysics analysis. To validate the proposed multiphysics model, experimental results from the three-electrode system were employed. Based on the measured conditions, the key parameters used in the model are summarized in SI Appendix, Table S2. Under the condition of 3 ppm iron ions, the simulated polarization curves derived from the model exhibit good agreement with the experimental data, confirming the accuracy of the model (SI Appendix, Fig. S11).

Fig. 3.

Multiphysics mechanism elucidation of magnetic enhancement and parameter influence analysis. (A) Schematic of the three-dimensional simulation model. (B) Distortion of current lines (arrowed lines) and current density distribution in the electrolyte (contour). (C) X-direction Lorentz force distribution. (D) Flow streamlines (arrowed lines) and the x-direction flow velocity distribution (contour). (E) Comparison of the iron ion concentration distribution with and without magnetic field. (F) Simulated OER potential (vs. RHE) and surface iron concentration as a function of magnetic flux density. (G) Simulated OER potential (vs. RHE) with and without magnetic field as a function of bulk iron concentration. Note: (B–G) are simulated under an applied average current density boundary condition of 400 mA/cm².

As shown in Fig. 3B, the current lines are predominantly oriented perpendicular to the anode surface across most of the domain. However, the current lines undergo significant distortion at the anode edges due to the edge effect, leading to an increase in the parallel component to the anode surface. Meanwhile, this effect results in higher current density at the edges, with the maximum and minimum values on the anode being 832 mA/cm² and 247 mA/cm², respectively. Since the current boundary condition applied to the anode surface is 0.4 A, the average current density is 400 mA/cm². The magnitude of the Lorentz force is determined by both the magnetic flux density and the current density, with the direction following the left-hand rule. As shown in Fig. 3C, a large Lorentz force in the x-direction is generated at the edge of the anode, with the maximum x-component reaching 1,017 N/m³. At the anode’s upper edge, the Lorentz force is directed along the positive x-axis, whereas the direction reverses at the lower edge.

Subsequently, a stirring flow is induced by the Lorentz force in the electrolyte above the anode, with the flow rate distribution shown in Fig. 3D. The maximum x-direction flow velocity is 0.032 m/s. The centrifugal force generated by this stirring effect drives the electrolyte at the anode surface toward the edges, replenishing fresh electrolyte above the anode. Consequently, a secondary flow (46) distribution directed vertically toward the anode surface is formed, as indicated by the arrowed lines in Fig. 3D. To further analyze the magnetic modulation on dissolved iron, the iron ion concentration distributions with and without the magnetic field are compared, as illustrated in Fig. 3E. In the absence of the magnetic field, iron ions tend to drift away from the anode surface under the influence of electric field. The concentration at the anode surface is as low as 0.072 ppm. Integration over the anode surface gives the activation overpotential of 0.293 V without magnetic field. According to NP equation, the concentration distribution of iron ions is determined by the contributions of diffusion, electromigration, and convection. The magnetic field-induced disturbances in the electrolyte, particularly the flow directed toward the anode, lead to a convective flux of iron ions that significantly exceeds both the electromigration and diffusion fluxes. Therefore, the iron ions are confined to the anode surface with a minimum concentration of 1.64 ppm, as shown in Fig. 3E. The activation overpotential at the anode surface is then reduced to 0.245 V.

Moreover, parametric analysis was performed to identify suitable magnetic design parameters, providing practical guidance for magnet-performance matching. Fig. 3F illustrates the effect of varying magnetic flux density on the anode potential and surface iron ion concentration at a constant current density of 400 mA/cm². To better visualize the details, x-axis range from 0 to 0.05 T was stretched. As the magnetic flux density increases from 0, the iron ion concentration at the anode surface exhibits a two-stage trend: a rapid initial increase, followed by a more gradual rise after 0.05 T. The magnetic flux density ranges corresponding to the two stages are defined as the field-sensitive region and the field-saturated region. Similarly, the anode potential follows an inverse trend, which can be explained by the modified BV equation. The voltage drop of 41 mV is observed from 0 to 0.05 T, while only 4 mV further reduction occurs between 0.05 and 0.1 T, indicating the magnetic saturation. In the field-sensitive region, the induced fluid disturbances enhance iron ion transport toward the anode surface, accelerating the accumulation of iron species. However, as the magnetic flux density further increases to the field-saturated region, iron ion transport becomes constrained by the concentration limit in the bulk electrolyte. Once the surface iron ion concentration approaches this upper bound, the accumulation rate slows down. The findings indicate that continuously increasing the magnetic flux density does not necessarily yield greater benefits. Considering the economic cost of magnetic sources, the optimal magnetic flux density appears to be near the transition point (0.05 T) between the two regions.

Fig. 3G quantifies the relationship between bulk iron concentration and anode potential at the current density of 400 mA/cm². Overall, the anode potential decreases with increasing bulk iron concentration, which aligns with the modified BV equation. Additionally, the potential difference between the two conditions increases as bulk concentration rises. This trend can be attributed to the concentration-dependent effect of magnetic regulation. At lower concentrations, the enhancement induced by the magnetic field is relatively limited, as the surrounding iron ion supply constrains the achievable surface accumulation. Consequently, the performance improvement remains modest in this region. However, as the bulk iron concentration increases, the magnetic field more effectively facilitates iron ion transport and accumulation at the anode surface, leading to a greater reduction in overpotential. As illustrated in Fig. 3G, to achieve the anode overpotential of 1.523 V at 400 mA/cm², the bulk iron concentration of 3 ppm is required in the absence of a magnetic field. In contrast, the same overpotential is achieved with only 0.3 ppm under the 0.18 T magnetic field. This indicates that the localized modulation effect of the magnetic field is equivalent to a tenfold increase in the bulk iron concentration, thereby effectively enhancing the OER activity.

Industry-Scale Application and Techno-Economic Analysis of ME-AWE.

Building on the above analysis, magnetic field applied perpendicular to the anode effectively increases the coverage of dynamic iron active sites and enhances reaction performance. Under industrial conditions, trace iron ions inevitably leach from steel pipelines and bipolar plates into the electrolyte. To further utilize the proposed ME-AWE technology under high current density, high temperature and industrial flow channel conditions, the AWE electrolyzer with the anode-side vertical magnetic field is designed and experimentally evaluated through the industrial BoP system.

The ME-AWE device, developed based on industrial processes integration, primarily consists of the measurement and control module, gas–liquid separator, heat exchanger, circulation pump, and electrolyzer, which serves as the core of the apparatus (Fig. 4 A and B). The electrolyzer features a split-type bipolar plate configuration, where the anode flow module is separately fabricated from axially magnetizable neodymium-iron-boron (NdFeB), while the remaining part of the bipolar plate is made of carbon steel (Fig. 4C). To prevent magnet demagnetization under industrial operation at 80 °C, the N40M-grade NdFeB magnet with a 100 °C tolerance is selected, with detailed parameters in SI Appendix, Table S3. The electrolyzer consists of six cells connected by the split-type magnetizable bipolar plate structure, and each cell contains two electrodes, separated by the membrane and insulation layer (Fig. 4D). For comparative analysis, Cell 2 is left unmagnetized (NM-AWE cell) and Cell 4 is axially magnetized (ME-AWE cell), with the two cells arranged nonadjacently to minimize magnetic interference.

Fig. 4.

Implement and performance of the industrial ME-AWE device. (A) Photograph of the industrial ME-AWE device. (B) Enlarged view of the assembled industrial electrolyzer for comparison of ME-AWE and NM-AWE. (C) Split-type magnetizable bipolar plate on the anode side. (D) Schematic illustration of the structure of the magnetic cell and nonmagnetic cell. (E) Cell voltages of the ME-AWE and NM-AWE at constant current density of 400 mA/cm² for over 500 h. (F) Comparison of voltage-current density curves between ME-AWE and NM-AWE at 80 °C, 70 °C and 60 °C. (G) Reduction in power consumption enabled by magnetic enhancement across various current densities and temperatures.

To evaluate the performance advantages and stability of ME-AWE in industrial applications, a long-term operation test exceeding 500 h is conducted at 400 mA/cm² and 80 °C (Fig. 4E). Notably, the ME-AWE cell exhibits a faster initial voltage drop during activation, leading to a progressively increasing voltage advantage that stabilized at 45 mV, consistent with the CV accelerated activation trend observed in the three-electrode tests (Fig. 2B). During the operation, ME-AWE cell sustained a persistently reduced voltage and exhibited a lower voltage degradation rate (~26 μV/h) than the NM-AWE (~44 μV/h). The ME-AWE ultimately achieved a 52 mV voltage advantage, indicating that the electricity saved is about 1.4 kWh to generate each kilogram of hydrogen gas. Postoperation EPMA analysis of the anodes further reveals a substantial increase in surface iron content for ME-AWE relative to NM-AWE (6.45 wt% vs. 3.25 wt%), providing quantitative evidence that the proposed magnetically driven mechanism remains effective in industrial-scale electrolyzer operation (SI Appendix, Figs. S12 and S13).

The voltage-current density curves at different temperatures (Fig. 4F) exhibit the consistent magnetically induced voltage reduction, further validating the performance enhancement enabled by ME-AWE. In constant current mode, reducing the cell voltage is a fundamental strategy for improving efficiency and reducing energy consumption. The ME-AWE cell demonstrates a substantial voltage reduction across the industrial current density range of 100 to 480 mA/cm² at 80 °C. At the rated current density of 400 mA/cm², the cell voltage is reduced by 43 mV, corresponding to a 2.4% improvement in hydrogen production efficiency. Given a fixed reaction area, the hydrogen production rate scales linearly with current density. Accordingly, the application of the magnetic field increases the current density from 327 mA/cm² to 408 mA/cm² at 1.8 V, corresponding to a 24.9% enhancement in hydrogen output. Notably, the minimal change in slope within the ohmic region (47) indicates a negligible magnetic improvement on bubble effect (48), consistent with the results in the previous three-electrode experiments (Fig. 2A). From the perspective of electrolysis energy consumption, the reduction rate is more pronounced at low current densities and elevated temperatures (Fig. 4G). Notably, since the magnetic modulation targets the intrinsic activity of the anode material, effectiveness is maintained even in the low-current-density region of the characteristic curve, which is unattainable by temperature increase alone. For industrial AWE devices, this translates to a reduction in the minimum operating power, effectively expanding the dynamic working range of the electrolyzer and improving the compatibility with intermittent renewable energy sources. From the hydrogen production perspective, the enhanced performance under low-load conditions also enables greater hydrogen output at partial loads, which is critical for maximizing system efficiency during periods of limited power availability. In terms of renewable energy input, this effect demonstrates the potential to improve integration and utilization of renewables within coupled hydrogen production systems.

To assess the practical potential of ME-AWE, a life cycle techno-economic analysis is performed under scenarios involving an off-grid green hydrogen production plant (OGHP) at the 10⁴ t/y scale (Fig. 5A), using the levelized cost of hydrogen (LCOH) as the key metric (38, 49). Overall, ME-AWE requires fewer electrolyzer units and achieves lower LCOH compared to NM-AWE across different wind capacity configurations that meet the annual hydrogen production target (Fig. 5B). For the onshore wind scenario, a minimum wind capacity of 180 MW is required to meet the annual hydrogen production target of 10⁴ t. Due to the higher electrolysis efficiency and lower operating threshold of the ME-AWE, 33 units of 1,000 Nm³/h devices are needed, which is five fewer than in the NM-AWE configuration. This reduction in electrolyzer count significantly lowers the overall capital investment for the OGHP, while achieving a comparable hydrogen yield. As a result, LCOH under the ME-AWE configuration can be as low as 2.03 $/kg, representing a 0.07 $/kg reduction compared to NM-AWE. By 2030, global production of low-emission hydrogen is projected to reach 6.5 × 10⁷ t/y, according to the IEA (5). With this figure, the total cost of hydrogen production is estimated to be reduced by 4.55 × 10⁹ $/y. For the offshore wind scenario, the higher annual full-load hours reduce the required wind capacity. Although the overall LCOH is elevated due to the higher cost of offshore wind turbines, the ME-AWE configuration consistently results in fewer electrolyzer units and thereby a lower LCOH compared to NM-AWE. These findings confirm that the proposed industrially validated ME-AWE can effectively enhance the techno-economic performance of OGHPs across diverse deployment scenarios, highlighting its adaptability and application potential.

Fig. 5.

Techno-economic analysis of the renewable-powered hydrogen production system integrated with ME-AWE. (A) Schematic of off-grid green hydrogen production plant (OGHP). (B) LCOH under different electrolyzer and wind power capacity deployment schemes for large-scale hydrogen production (10⁴ t/y level) based on offshore and onshore wind power. (C and D) Optimized hourly electrolysis power over a full year under (C) NM-AWE and (D) ME-AWE conditions with 180 MW onshore wind power. (E) Cost breakdown comparison of AWE devices between NM-AWE and ME-AWE configurations. (F) LCOH breakdown under ME-AWE configuration.

Subsequently, a detailed analysis is conducted based on the 180 MW onshore wind power configuration. According to the 8,760-hourly optimized operation results (Fig. 5 C and D), under the same wind power capacity, ME-AWE exhibits a lower peak electrolysis power due to fewer installed AWE units, leading to the higher wind curtailment rate of 1.42%. Nevertheless, owing to the enhanced electrolysis performance, the ME-AWE-based OGHP still meets the 10⁴ t/y hydrogen production target and even exceeds that of NM-AWE by 39 t. From the perspective of AWE devices utilization, the ME-AWE configuration increases the annual full-load hours of the hydrogen production to 2,743 h, representing a 13.5% improvement. Regarding the total cost of AWE devices, the ME-AWE configuration amounts to 54.0 million $, representing a 10.6% reduction compared to NM-AWE (Fig. 5E). The operation and maintenance (O&M) cost, mainly including consumables associated with hydrogen produced and maintenance expenses tied to total device capacity, is reduced by 0.24 million $ in the ME-AWE configuration. In terms of AWE’s capital investment (stack and BoP), the magnetic component in ME-AWE slightly increases the stack cost. However, this accounts for only 3% of the total investment and is offset by the substantial reduction in the number of AWE units. The nonmagnetic portion of the stack and BoP, which dominate the total cost and scale directly with the number of units, are significantly reduced, effectively compensating for the marginal added cost of the magnetic material and reducing the LCOH. From the LCOH breakdown of the ME-AWE-configured OGHP (Fig. 5F), the wind power plant accounts for the largest share of the total cost, reaching 1.401 $/kg and representing 68.9% of the LCOH, primarily due to the relatively high unit investment cost of wind power. While the increased AWE deployment in the NM-AWE configuration reduces the proportional contribution of wind power cost, it remains the primary cost driver, representing 66.8% of the LCOH (SI Appendix, Fig. S14). In the long term, the LCOH of OGHP is expected to be further reduced with falling wind power cost. Therefore, considering the superior techno-economic performance and compatibility with existing industrial processes, the proposed ME-AWE demonstrates the potential for large-scale green hydrogen production from renewable energy.

Discussion

This study proposes an ME-AWE strategy to boost OER activity by modulating trace iron ions in the electrolyte, enabling a scalable transition from laboratory investigation to industrial application. By inducing Lorentz-force-driven convection, the magnetic field promotes local enrichment of iron ions near the anode, increasing the coverage of dynamic iron active sites and enhancing catalytic performance. This mechanism is first validated through controlled three-electrode experiments combined with comprehensive characterizations. To quantitatively interpret the observed enhancement, a multiphysics model coupling magnetic field, electrochemical reaction, fluid flow, and mass transport is developed. Lumped-parameter simulation reveals that the enhancement achieved through magnetic modulation at 0.3 ppm iron matches that of tenfold bulk iron concentration under nonmagnetic conditions. Moreover, the magnetic effect exhibits saturation above a critical flux density (~0.05 T), offering practical guidance for the optimized application of magnetic fields in balancing performance gains and material costs. At the industrial scale, the ME-AWE strategy is deployed using commercially relevant electrolyzer architectures without altering core process infrastructure, exhibiting a 24.9% increase in hydrogen production rate at 1.8 V under typical operating conditions (80 °C). Ultimately, under the optimized operation of various configurations within the 10⁴ t/y OGHP scenario, ME-AWE consistently demonstrates the capability to reduce the number of required AWE units and improve their utilization. Although the introduction of the magnetic components slightly increases the stack cost, this is effectively offset by the downsized AWE deployment, resulting in a lower overall LCOH of the OGHP. This further underscores the techno-economic advantages and practical feasibility of the ME-AWE strategy for large-scale green hydrogen production.

Beyond immediate device-level benefits, the findings also open avenues for leveraging electrolyte impurities under magnetic modulation, offering design insights for future impurity-assisted strategies and nonprecious-metal catalyst development. Through the strategic integration of high-temperature-resistant permanent magnets or electromagnets, the desired enhancement can be achieved in a controllable and sustainable manner with minimal carbon emissions. Looking forward, advances in magnetic field design, magnet recyclability, and the coordinated development of impurity-assisted materials are anticipated to readily elevate the impact of magnetic enhancement in AWE. The insights play a vital role in advancing green hydrogen production technologies from laboratory to industry and accelerating the global transition to sustainable energy.

Methods

Electrochemistry Studies.

Materials.

The working electrode (WE) employed a commercial nickel foam sheet (1 cm × 1 cm), while a platinum sheet (1 cm × 1 cm) served as the counter electrode (CE). Ag/AgCl electrode was used as the reference electrode (RE) to ensure stable potential measurements. Commercial nickel foam was first immersed in 1 M HCl solution and subjected to ultrasonication for 10 min at room temperature, followed by thorough rinsing with deionized water. An axially magnetized cylindrical NdFeB permanent magnet (N40, Φ7.2 cm × 1.8 cm) was used as the magnetic field source. The magnetic flux density at the anode surface was measured at 0.18 T using the Gauss meter (Sanliang, TS500). Given the small electrode area and the proximity to the magnet’s central axis, the magnetic field at the electrode surface could be approximated as uniform in both magnitude and direction. To conduct accelerated testing under industrially relevant conditions, a trace iron ion concentration of 3 ppm (equivalent to 0.054 mol/m³) was introduced into the 30 wt% KOH electrolyte.

Synthesis of NiFe electrode.

0.9 mmol Ni(NO₃)₂·6H₂O, 0.3 mmol Fe(NO₃)₃·9H₂O and 3.6 mmol urea were dissolved in 30 mL deionized water under vigorous stirring, followed by the immersion of a prewashed nickel foam into the solution and reaction in a Teflon-lined autoclave at 120 °C for 12 h. After washing with deionized water and drying in a vacuum oven at 25 °C for 12 h, the NiFe catalyst was obtained.

Electrochemical measurements.

The electrochemical workstation (CH660E) was employed to characterize the OER kinetics under the controlled experimental conditions. Each group of experiments was preconditioned with 40 consecutive cyclic voltammetry (CV) cycles to accelerate electrode activation and electrochemical stability. To verify the accuracy of electrochemical measurements, the open-circuit potential was recorded prior to each test to assess any deviation between the in-use Ag/AgCl reference electrode and a brand-new one. All current densities in this paper were based on the apparent geometric area of the electrodes. The potential vs. RHE could be calculated as follows:

$$\mathrm{E}(\mathrm{RHE})=\mathrm{E}(\mathrm{Ag}/\mathrm{AgCl})+0.197+0.059\times \mathrm{pH}.$$ [1]

CV was performed at a scan rate of 5 mV/s within a potential window of 0 to 0.8 V vs. Ag/AgCl. Linear sweep voltammetry (LSV) was also recorded at the scan rate of 5 mV/s. Electrochemical impedance spectroscopy (EIS) was conducted over a frequency range of 10⁴ to 10⁻¹ Hz. To assess the intrinsic electrocatalytic activity, the solution resistance (R_s) was used for IR compensation in LSV measurements. All electrochemical tests were carried out at room temperature.

Characterizations.

XPS was performed using a Thermo Scientific K-Alpha instrument with C 1 s (binding energy of 284.6 eV) as the reference. TEM image and element mapping were acquired using an FEI Talos F200S. EPMA-WDS was carried out with an EPMA-8050G instrument.

Multiphysics Model.

Current-magnetic field coupled model.

The magnetic field distribution is governed by Maxwell’s equations. In the present steady-state simulation, displacement currents, and time-varying fields are neglected and the magnetic field is assumed to be entirely generated by the stationary permanent magnet. Therefore, the magnetic flux density B in the electrolyte satisfies the divergence-free condition:

$$\nabla ·\mathit{B}=0,$$ [2]

and is related to the magnetic scalar potential φ_m by

$$\mathit{B}=-{\mu }_0{\mu }_{r,l}\nabla {\phi }_m,$$ [3]

where μ₀ represents the magnetic permeability of vacuum, μ₀ = 4π × 10⁻⁷ H/m. μ_r,l is the relative magnetic permeability of the electrolyte.

The electrolyte is treated as an electrically neutral conductive fluid with a certain conductivity. In the presence of the magnetic field, the motion of the electrolyte gives rise to an induced electric field. According to the Laplace equation, electric potential within the electrolyte and the electrode can be respectively described as follows:

$$\nabla ·({\sigma }_l\nabla {\phi }_l+\mathit{u}\times \mathit{B})=0,$$ [4]

$$\nabla ·({\sigma }_s\nabla {\phi }_s)=0,$$ [5]

where σ_l and φ_l are the conductivity and the electric potential of the electrolyte. u stands for the flow velocity. σ_s and φ_s are the conductivity and the electric potential of the electrode.

The voltage of the anodic half-cell E_a in the AWE electrolyzer can be modeled as follows:

$$E_a=E_{rev,a}+{\eta }_{ohm,a}+{\eta }_a,$$ [6]

where E_rev,a is the anodic reversible voltage, representing the minimum voltage at which the OER can occur. η_ohm is the ohmic overpotential. η_a is the anodic activation overpotential, indicating the electrochemical polarization at the anodic electrolyte–electrode interface.

According to the dynamic iron active site theory (35), the iron content at the anode is proportional to the iron ion concentration in the electrolyte at the anode–electrolyte interface. To reflect this effect on the electrochemical reaction, the coverage factor θ_Fe is introduced to represent the fraction of the anode surface covered by dynamic iron active sites. The remaining surface is assumed to retain the properties of the original electrode material. Therefore, the relationship between the anodic activation overpotential and current density can be expressed by the modified BV equation:

$$j_r=j_{0,a}^{\mathit{eq}}\left(exp\left(\frac{{\alpha }_{a}F{\eta }_a}{\mathit{RT}}\right)-exp\left(\frac{-{\alpha }_{c}F{\eta }_a}{\mathit{RT}}\right)\right),$$ [7]

$$j_{0,a}^{\mathit{eq}}=j_{0,a}^{\ast }{\theta }_{\mathit{Fe}}+j_{0,a}\left(1-{\theta }_{\mathit{Fe}}\right),$$ [8]

$${\theta }_{\mathit{Fe}}=\frac{c_{\mathit{Fe}}^0}{c_{\mathit{Fe}}^{\ast }},$$ [9]

where j_r represents the reaction current density. $j_{o\mathit{,}a}^{eq}$ is the equivalent exchange current density of the anode. $j_{0,a}^{\ast }$ is the exchange current density at maximum stable iron coverage on the anode and $c_{\mathit{Fe}}^{\ast }$ is the corresponding iron ion concentration at the anode surface. j_0,a is the exchange current density without iron on the anode. $c_{\mathit{Fe}}^0$ represents the actual surface iron ion concentration at the anode. α_a and α_c denote the transfer coefficient of the anode and cathode respectively. R is the ideal gas constant. T is the reaction temperature.

The boundary conditions are as follows:

(1) The total current at the anode surface is defined as a constant I_total:

$${\iint }_{\mathit{surface}}{\mathit{j}}_{\mathit{r}}·\mathit{n}dA=I_{\mathit{total}},$$ [10]

where n is the normal vector of anode surface.

(2) Considering the symmetry of the electric field, the potential at the boundary of the symmetrical plane between the anode and cathode is set as a constant φ_ref:

$${\phi }_l={\phi }_{\mathit{ref}}.$$ [11]

(3) Other boundaries are defined as insulation:

$$-{\mathit{j}}_{\mathit{l}}·\mathit{n}=0.$$ [12]

Flow-mass transport coupled model.

As discussed above, the local concentration of dissolved iron ions at the anode surface plays a critical role in governing the OER kinetics during water electrolysis. To precisely describe the transport behavior, a coupled model integrating fluid dynamics and mass transport is developed.

The transport of dissolved iron ions in the electrolyte is governed by a mass conservation equation in the form of a flux divergence:

$$\nabla ·{\mathit{N}}_{\mathit{Fe}}=R_{\mathit{Fe}},$$ [13]

where N_Fe is the total iron ion flux. R_Fe stands for the volumetric source term. In this study, iron ions are treated as dynamically stable species that modulate catalytic active sites without being consumed or produced via Faradaic reactions. Accordingly, there is no net generation or depletion of iron species in the electrolyte and the source term R_Fe is set to zero.

N_Fe is described by the NP equation, which incorporates diffusion, electromigration, and convection effects:

$${\mathit{N}}_{\mathit{Fe}}=-D\nabla c_{\mathit{Fe}}-z_{\mathit{Fe}}{u}_{m,Fe}F{c}_{\mathit{Fe}}\nabla {\phi }_l+c_{\mathit{Fe}}\mathit{u},$$ [14]

where D, c_Fe, z_Fe, u_m,Fe represent the diffusion coefficient, concentration, charge number, and electromobility of iron ions, respectively.

When applying the static magnetic field, the flow dynamics of the electrolyte above the anode is governed by the continuity equation and the Navier–Stokes equation:

$$\nabla ·\mathit{u}=0,$$ [15]

$$\rho (\mathit{u}·\nabla )\mathit{u}=-\nabla p+{\mu }_{\mathit{vis}}{\nabla }^2\mathit{u}+{\mathit{F}}_{\mathit{L}},$$ [16]

where ρ, p, μ_vis, represent the density, pressure, and viscosity of the electrolyte, respectively. F_Ldenotes the volumetric Lorentz force generated by the electromagnetic field and is defined as

$${\mathit{F}}_{\mathit{L}}={\mathit{j}}_{\mathit{l}}\times \mathit{B}.$$ [17]

The boundary conditions for the concentration and flow field:

(1) No iron ion flux at the anode surface:

$$-\mathit{n}·{\mathit{N}}_{\mathit{Fe}}=0.$$ [18]

(2) The iron ion concentration on other walls is defined as a constant c_Fe,bulk:

$$c_{\mathit{Fe}}=c_{Fe,bulk}.$$ [19]

(3) No-slip boundary condition at the anode surface:

$$\mathit{u}=0.$$ [20]

(4) Open boundaries on other walls:

$$p=0.$$ [21]

Industry-Scale AWE Implementation of Magnetic Field.

The AWE electrolyzer comprised six individual electrolytic cells, each with an effective reaction area of 20 cm², utilizing commercial nickel foam electrodes for both the anode and cathode. Each cell of the electrolyzer was designed with a split-type magnetizable configuration (Fig. 4 C and D). Cell 4 was selected for axial magnetization (residual magnetization: 1.267 T), while Cell 2 remained unmagnetized for comparison (Fig. 4 B and D). A 30 wt% KOH was employed as electrolyte in the experiment. Additionally, the industrial BoP system was integrated, consisting of the heat exchanger, circulation pump, gas–liquid separator, and measurement and control module. During operation, the electrolytic current could be adjusted via the control module and the electrolyte inlet velocity is set at 0.5 m/s. The voltage of each electrolytic cell was recorded at the sampling rate of 1 S/s.

Optimization Model of the OGHP.

The optimization model of the OGHP is established to determine the optimal 8,760-hourly operation of the installed AWE units under a fixed wind power capacity, aiming to maximize annual hydrogen production. The electrolysis characteristics are represented using the piece-wise linear (PWL) function (50) based on experimentally derived performance curves from industry-scale ME-AWE and NM-AWE. The LCOH is directly obtained by integrating the optimized hourly operation with lifetime assumptions, enabling lifecycle techno-economic assessment.

The techno-economic model was developed under typical industrial assumptions. Both ME-AWE and NM-AWE systems are assumed to have a 20-year operational lifetime, consistent with current MW-scale alkaline electrolyzers. The embedded NdFeB permanent magnets (grade N40M) have a maximum working temperature of 100 °C, which ensures stable magnetic performance at the rated operating temperature of 80 °C and thus negligible demagnetization over the stack lifetime. Therefore, the magnetic components are treated as a capital cost added to the stack investment without considering replacement.

The objective function is defined in Eq. 22, with constraints detailed in Eqs. 23–25. The modeling framework of the optimization model for OGHP is described in SI Appendix, Note 1. Moreover, the per-unit wind power profile and all model parameters are provided in SI Appendix, Fig. S15 and Table S4. The objective function (Eq. 22) is equivalently formulated to maximize the annual hydrogen production. The decision variables include the 8,760-hourly on/off status, operating power, and hydrogen production rate of the electrolyzer units.

$$J\text{=}min\left(-{\displaystyle \sum _{t=1}^T}{\displaystyle \sum _{i=1}^M}{Q}_{i,t}\right),$$ [22]

where t is the hourly time index, i is the index of electrolyzer, T is the total number of hours in a year (8,760) and M is the total number of installed electrolyzer units. Q represents the hydrogen production rate.

The constraints include power balance in Eq. 23, the lower and upper power limits of the electrolyzer in Eq. 24 and power-to-hydrogen conversion relationship in Eq. 25.

$$P^{WP,r}·P_t^{WP,pu}={\displaystyle \sum _{i=1}^M}{P}_{i,t}^{\mathit{ELE}}+P^{WP,r}·\mathrm{\Delta }{P}_t^{WP,pu},$$ [23]

$$P_{min,i}^{\mathit{ELE}}·U_{i,t}^{\mathit{ELE}}\le P_{i,t}^{\mathit{ELE}}\le P_{max,i}^{\mathit{ELE}}·U_{i,t}^{\mathit{ELE}},$$ [24]

$$Q_{i,t}={\displaystyle \sum _{k=1}^K}{a}_{i,k}{P}_{i,t,k}^{\mathit{ELE}}+b_{i,k},$$ [25]

where P^WP,r is the rated power of the wind power plant. $P_t^{WP,pu}$ is the per-unit wind power output at time t. $\mathrm{\Delta }{P}_t^{WP,pu}$ denotes the curtailed per-unit wind power at time t. $P_{i,t}^{\mathit{ELE}}$ is the power of the electrolyzer i at time t. $P_{min,i}^{\mathit{ELE}}$ and $P_{max,i}^{\mathit{ELE}}$ are the minimum and maximum operating powers of the electrolyzer i. $U_{i,t}^{\mathit{ELE}}$ is the on/off status (binary variable) of the electrolyzer i at time t. k is the segments index in the PWL function of the energy conversion characteristics. a_i,k and b_i,k are the slope and intercept of the segment k for the electrolyzer i. $P_{i,t,k}^{\mathit{ELE}}$ denotes the power assigned to segment k for the electrolyzer i at time t.

Definition of LCOH.

The LCOH represents the average lifecycle cost per unit mass of hydrogen and serves as an effective indicator for evaluating the economics and competitiveness of various routes. The LCOH for the OGHP can be calculated as follows:

$$LCOH=\frac{C_{\mathit{inv}}+C_{\mathit{om}}}{{\sum }_{n=1}^N\frac{H_{\mathit{output}}}{{\left(1+r\right)}^n}},$$ [26]

$$H_{\mathit{output}}={\displaystyle \sum _{t=1}^T}{\displaystyle \sum _{i=1}^M}{Q}_{i,t},$$ [27]

$$C_{\mathit{inv}}=c_{inv,WP}·I{C}_{\mathit{WP}}+c_{inv,ELE}·I{C}_{\mathit{ELE}},$$ [28]

$$C_{\mathit{om}}={\displaystyle \sum _{n=1}^N}\frac{c_{om,WP}·I{C}_{\mathit{WP}}+c_{w,ELE}·{\mu }_{w,ELE}·H_{\mathit{output}}+c_{m,ELE}·I{C}_{\mathit{ELE}}}{{\left(1+r\right)}^n},$$ [29]

where C_inv represents the total investment cost. C_om is the total operation cost. r is discount rate. N is the lifespan of the complete project. H_output denotes the annual produced hydrogen mass. c_inv,WP and c_inv,ELE are the cost of unit investment of wind turbine and electrolyzer respectively. IC_WP and IC_ELE are the capacity of the wind power and electrolyzer respectively. c_om,WP is the costs of unit operation and maintenance of wind power. c_w,ELE and μ_w,ELE are the cost of unit water and water consumption coefficient. c_m,ELE represents the cost of unit maintenance of the electrolyzer.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

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