Achieving volatile potassium promoted ammonia synthesis via mechanochemistry

Abstract

Potassium oxide (K₂O) is used as a promotor in industrial ammonia synthesis, although metallic potassium (K) is better in theory. The reason K₂O is used is because metallic K, which volatilizes around 400 °C, separates from the catalyst in the harsh ammonia synthesis conditions of the Haber-Bosch process. To maximize the efficiency of ammonia synthesis, using metallic K with low temperature reaction below 400 °C is prerequisite. Here, we synthesize ammonia using metallic K and Fe as a catalyst via mechanochemical process near ambient conditions (45 °C, 1 bar). The final ammonia concentration reaches as high as 94.5 vol%, which was extraordinarily higher than that of the Haber-Bosch process (25.0 vol%, 450 °C, 200 bar) and our previous work (82.5 vol%, 45 °C, 1 bar).

Fig. 1. Comparison of Fe catalytic activity with different K promotors.

a Schematic of Fe catalytic activity with different promotors depending on reaction temperature. Pure K and oxygenated K derivatives (K₂O and KOH, both denoted as K + O) are used as K promotors for NH₃ synthesis. At a temperature above 400 °C, the activity of K on Fe (FeK) is diminished by the K desorption from the surface of Fe catalyst. Therefore, oxygenated K derivatives on Fe [Fe(K + O)s] show better performance above 400 °C. However, at a temperature below 400 °C, the K can stably stick to the surface of FeK, showing much higher activity than Fe(K + O)s. Color code in the cartoon: gray, dark red, and purple circles are Fe, K + O, and K, respectively. b Comparison of synthesized NH₃ concentration with same catalyst weight, using different K promotors (K, K₂O, and KOH) at the same conditions [same loading ratio of K to Fe (2.5 at%), 45 °C] via mechanochemical process. For N₂ dissociation step, ball-milled for 11.5 h after charging N₂ (9 bar). For hydrogenation step, ball-milled for 3 h after charging H₂ (9 bar). The error bars are determined from at least five independent experiments. Density functional theory calculations of mechanochemical ammonia synthesis with different catalytic systems (FeK, FeK₂O, and Fe): c nitrogen dissociation step; d hydrogenation step.

Fig. 2. N₂ dissociation kinetics.

a The amount of adsorbed N₂ with respect to rotation speed. The total rotating cycles for FeK were 108,000 cycles, and for pure Fe were 240,000 cycles. Data were normalized to 30,000 cycles (1 hour at 500 r.p.m.). The data of Fe were adapted from ref. . b Highest N* fixation activity comparison between FeK and Fe. N* fixation yield was normalized by consumed energy (kWh) (FeK 400 r.p.m., Fe 450 r.p.m.). The data of Fe were adapted from ref. . The error bars (Fig. 2a, b) are determined from at least five independent experiments. c The amount of adsorbed N₂ with respect to ball-milling time. The rotation speed was 400 r.p.m. for both FeK and pure Fe. d The amount of adsorbed N₂ and the natural logarithm of the ball-milling time [ln(time)] are linearly dependent. Incubation period (FeK, 1.1 h; pure Fe, 6.0 h) could be determined by extrapolation to the origin. The weight of FeK and Fe was 23.8 g and 24 g.

Fig. 3. Ammonia yield kinetics.

a The amount of synthesized NH₃ versus rotation speed. The N₂ absorbed FeK powders were prepared by ball-milling in N₂ pressure (9 bar) for 18 h. The total number of rotation cycles for FeK were 60,000, and for N₂ absorbed Fe were 120,000. Data were normalized to 30,000 cycles (at 500 r.p.m. for 1 h). b The amount of synthesized NH₃ versus time, which is the hydrogenation rate at the rotation speed of 500 r.p.m. for both FeK and Fe. Hydrogenation was repeated after charging H₂ pressure (9 bar each) every 2 h for FeK, and every 4 h for Fe. c Relation between the natural logarithm of the synthesized NH₃ vs. the natural logarithm of time shows a linear dependence for FeK. d Final NH₃ concentration with respect to initially charged H₂ pressure. Each sample was ball-milled for hydrogenation for 2 h and 4 h after charging 9 bar of H₂. The weight of FeK and Fe was 23.8 g and 24 g. All the data of Fe were adapted from ref. . The error bars (Fig. 3a, d) are determined from at least five independent experiments.

Fig. 4. Theoretical study on the electronic effect between Fe and K.

a Partial density of states (PDOS) of Fe(110)-K, including the Fe-d orbits and adjacent K-s orbits. b Corresponding charge density difference between Fe and K. c the Hirshfeld charge analysis. Color codes in the cartoon: purple and gray are K and Fe, respectively. From the Hirshfeld charge analysis in b, the red regions represent electron accumulation and the blue regions represent electron depletion, respectively (Max Isovalue: 0.02 e Å⁻³).

Fig. 5. Characterizations of the catalysts.

a XRD measurements of FeKN*, and regenerated FeK after hydrogenation. b Radial distribution function of FeKN* and regenerated FeK after hydrogenation. Radial distribution function was fitted by FeK-edge EXAFS data. c Mössbauer spectra of FeKN*. High-resolution XPS N 1 s spectra: d FeKN* and e FeN*. f NEXAFS of FeKN*, FeN*, and commercial Fe_xN (x = 2–4).