Practical H₂ supply from ammonia borane enabled by amorphous iron domain

Hydrogen is an attractive, clean energy carrier, but practical, safe, and cost-effective storage and on-demand release remain major barriers for vehicular applications. Compressed hydrogen storage raises safety concerns, motivating chemical storage routes in which catalysts release H₂ from carriers such as ammonia borane (AB), which has high hydrogen content (19.6 wt%), air/water stability, and non-toxicity. However, AB hydrolysis catalysts typically lack sufficient durability (usually <2 h; a noble-metal benchmark Rh/Co₃O₄ reached 45 h but activity dropped sharply within 9 h), limiting practicality. Engineering micro/nanostructures can boost activity but often compromises stability, whereas bulk materials offer stability but low activity. The authors hypothesize that introducing amorphous domains with abundant defects and unsaturated atoms into a stable bulk Fe framework could couple high activity with long-term stability for efficient, durable AB hydrolysis and practical H₂ supply.

Prior work has explored a variety of catalysts for AB hydrolysis, including transition metal nanoparticles (e.g., Ni, Co, Ni₂P), supported catalysts, and noble-metal systems. While some systems show high initial activities, stability is generally limited to a few hours or cycles. A reported benchmark Rh/Co₃O₄ catalyst achieved a total of 45 h but with rapid deactivation after ~9 h. Amorphous domains in nanomaterials have been associated with enhanced catalytic activity due to abundant active sites and defect-rich environments, and defect engineering (e.g., oxygen vacancies, stacking faults) can activate otherwise less active materials. Fe-based nanomaterials can display some activity, but bulk Fe metal has not been reported as an efficient AB hydrolysis catalyst. DOE’s ultimate on-board H₂ storage target is 7.5 wt% H₂, and conventional AB aqueous solutions (~5 wt% H₂) often fall short without system-level optimization. These considerations motivate creating amorphous domains within bulk Fe to achieve both activity and durability, potentially enabling practical H₂ supply systems.

• Catalyst preparation: Commercial Fe foam (≥98.2% Fe; typically 1×1 cm², 1.6 mm thick) was cleaned (acetone/ethanol, sonication) and dried (60 °C, 5 h). Oxidation was performed in air (muffle furnace) at 800 °C for 7 h (also 300, 500, 1100 °C for comparison) to yield Fe₂O₃ Foam. Reduction was achieved by immersing the oxidized foam in 3 M AB + 0.5 M NaOH aqueous solution for 20 min at room temperature, followed by water/ethanol washing, producing R-Fe₂O₃ Foam. Alternative reductions included 3 M NaBH₄ or LiBH₄ (0.5 M NaOH, 20 min), hydrazine solution (reflux 95 °C, 12 h), and H₂ gas (10% H₂/Ar, 1100 °C, 3 h). Similar oxidation/reduction was applied to Ni, Co, and Cu foams for comparison.
• Structural/chemical characterization: SEM, TEM/HRTEM (FEI Talos F200X), aberration-corrected HAADF-STEM, XRD (PANalytical Empyrean PIXcel3D), XPS (Thermo K-Alpha+), ICP-OES, XAS (Fe K-edge XANES/EXAFS) at SSRF 11B and NSRL MCD-B, in situ Fe K-edge XANES/EXAFS during AB hydrolysis, B K-edge XANES, and positron annihilation lifetime spectroscopy (PALS) to probe defect states (τ₂, I₂ trends).
• Catalytic testing: Batch hydrolysis in a 25 mL one-neck round-bottom flask connected to a gas collection tube at room temperature and ambient pressure. Typical test used 2 mL solution containing 3 mmol AB and 0.5 mmol NaOH (NaOH-free shows slight performance drop). Reaction start time taken at first bubble; H₂ volume measured by water displacement; H₂ confirmed by GC. Area activity A = V_H2/(S_Foam × t). Long-term stability: a 1×1 cm² foam was cycled in a larger flask with 413 mL AB solution (9.52 g AB), refreshed every 24 h, with activity checked periodically in the 25 mL setup, over 900 h total. TOF calculated using surface-layer Fe as active sites based on BET area (2.3332 m²/g) and foam mass (0.192 g).
• Computational studies: DFT (VASP 5.3, GGA-PBE, PAW, 400 eV cutoff, CI-NEB and dimer methods for TS), models included Fe(110), FeB(111), Fe(110)-B, Fe(110) with vacancies (1V, 3V) with/without B, and an amorphous Fe₄₄ cluster with B (Fe₄₄-B). Adsorption energies and dissociation barriers for H₂O and NH₃BH₃ were computed; the rate-determining first O–H and B–H bond cleavages were emphasized. Simulated XRD confirmed amorphous character of Fe₄₄.
• Device demonstrations: On-board H₂ supply to a commercial fuel cell powering a model car; filtration water to reduce NH₃ in product H₂; tests of scaling by foam area including 10×10 cm² panels and arrays. GHSC and system energy density quantified, including scenarios using recycled fuel cell water.

• Performance: R-Fe₂O₃ Foam delivered TOF = 113.6 min⁻¹, continuous H₂ generation of ~771 L over 900 h without obvious decline, and area activity up to 43.27 mL/(min·cm²) for a 10×10 cm² panel. Catalytic activity scales linearly with foam area; 10×10 cm² foam produced >4000 mL H₂ per minute.
• Durability: Stable operation for at least 900 h; catalyst can work up to 1100 h until 20% of initial activity remains; reusability demonstrated with another ~300 h after re-oxidation/reduction treatment. In contrast, literature catalysts typically last <2 h; a Rh/Co₃O₄ benchmark totaled 45 h but deactivated sharply after 9 h.
• Practical H₂ supply: The system provided 180 mL H₂/min to a commercial fuel cell, stably powering a model car at ~7.8 V and 1.6 A (~12 W) for at least 5 h. The panel-like foam is easily removable to control reaction start/stop.
• Storage metrics: By storing AB as a solid and adding water stoichiometrically, GHSC reached 8.91 wt% H₂ (exceeding the DOE 7.5 wt% target). In a practical setup using recycled fuel cell water and ~19 wt% AB concentration, GHSC of 7.5 wt% H₂ was achieved. The onboard system attained specific energy densities of ~2.50 kWh/kg and 2.26 kWh/L, surpassing compressed H₂ at 350 bar (0.8 kWh/L) and 700 bar (1.3 kWh/L).
• Kinetics: Apparent activation energy Ea ≈ 20.74 kJ/mol; reaction rate exhibits zero-order dependence on AB concentration under tested conditions.
• Structure–function: TEM/HRTEM and HAADF-STEM revealed abundant amorphous Fe domains on a metallic Fe framework. XANES showed metallic Fe-like features; EXAFS indicated decreased coordination numbers consistent with vacancies/unsaturated atoms. PALS corroborated increased large defect populations (τ₂, I₂). Amorphous domains persisted after 500 h of operation.
• Mechanism and intermediates: In situ Fe K-edge XANES during AB hydrolysis showed spectral changes indicating formation of Fe–B intermediates; B K-edge XANES detected Fe–B features (surface sensitive). Spectral changes reverted after exposure to air, indicating Fe–B exists predominantly during operation.
• DFT insights: Models incorporating vacancies plus B (Fe(110)₃V–B and amorphous Fe₄₄–B) dramatically lowered dissociation barriers: Fe₄₄–B: H₂O 0.63 eV; AB dissociation barrierless with ΔE = –0.43 eV. Fe(110)₃V–B: H₂O 0.41 eV; AB 0.07 eV. By contrast, pure Fe(110): H₂O 1.35 eV; AB 1.13 eV; pure FeB(111): H₂O 1.13 eV; AB 0.81 eV. Vacancies enhance H₂O activation; B critically lowers AB activation; both are needed for optimal performance.
• Controls and comparisons: Pristine Fe foam and commercial Fe₂O₃ nanoparticles were inactive. Oxidation at ≥500 °C followed by AB reduction was necessary to create active amorphous domains; H₂ reduction at high temperature produced crystalline Fe without amorphous domains and poor activity; brief heating in air (60 °C) increased oxidation and deactivated the catalyst. B-containing reductants (NaBH₄, LiBH₄) effectively reduced Fe₂O₃ and yielded active catalysts; N₂H₄ reduction was ineffective. R-CoO showed high initial activity but poor stability (50% loss after 10 h); R-NiO and R-CuO had much lower activity than R-Fe₂O₃.

Introducing amorphous domains with abundant defects into a robust bulk Fe framework resolves the traditional trade-off between activity and durability for AB hydrolysis. The amorphous regions host vacancies and unsaturated Fe sites that can bind B under reaction conditions to form transient Fe–B intermediates, as supported by in situ XAS. DFT calculations reveal that vacancies and B synergistically and substantially reduce the activation barriers for the rate-determining O–H and B–H bond cleavages in H₂O and NH₃BH₃, respectively—barriers drop from >1 eV on crystalline Fe to ≤0.63 eV (and even barrierless for AB on amorphous Fe₄₄–B). Meanwhile, the metallic Fe skeleton maintains mechanical and chemical stability, preserving the amorphous domains for at least 900 h and enabling reactivation after re-treatment. The catalyst’s zero-order kinetics with respect to AB, low activation energy, and area-scalable performance support practical flow or panel-based reactor designs. Demonstrations of stable H₂ supply to a commercial fuel cell and high GHSC/system energy densities highlight relevance for on-board applications. Compared with prior catalysts that either lack durability or rely on costly noble metals, the low-cost Fe-based foam with engineered amorphous domains achieves orders-of-magnitude longer lifetimes while delivering high activity, indicating a viable pathway toward practical chemical hydrogen storage systems.

A simple oxidation–reduction strategy creates amorphous Fe domains within bulk Fe foam (R-Fe₂O₃ Foam) that serve as highly active sites for AB hydrolysis while the bulk crystalline Fe provides long-term stability. The catalyst delivers high activity (TOF 113.6 min⁻¹; area activity 43.27 mL/(min·cm²)), exceptional durability (≥900 h continuous operation; up to 1100 h until 20% initial activity), and practical H₂ supply to a commercial fuel cell, with system-level storage metrics meeting or exceeding DOE targets. Mechanistically, defect-rich amorphous domains form Fe–B intermediates under operation, dramatically lowering H₂O and AB dissociation barriers, as validated by in situ XAS and DFT. This work establishes a generalizable route to combine amorphous-domain reactivity with bulk structural robustness for durable, high-performance chemical hydrogen storage. Future work could optimize the density and distribution of amorphous domains, further elucidate transient Fe–B chemistry under operando conditions, integrate foam arrays for higher throughput, and extend the approach to other abundant metals and carriers alongside efficient regeneration of spent AB.

• The catalyst’s activity is sensitive to oxidation; heating in air at 60 °C increased oxidation and deactivated the catalyst.
• H₂ reduction at high temperature eliminates amorphous domains and activity, indicating performance depends critically on maintaining defect-rich amorphous regions.
• Alternative metal foams (e.g., Co, Ni, Cu) did not match the combined activity and stability of Fe; R-CoO exhibited rapid deactivation (≈50% activity loss after 10 h).
• Hydrazine reduction did not effectively reduce Fe₂O₃, and preparation relies on B-containing reductants to generate active amorphous domains.
• Demonstrations focused on model car and panel-scale tests; broader vehicular system validation and long-term operando impurity management (e.g., NH₃ removal) were not fully detailed.