Haber_process - Wikiparency

# Haber process  ## Microsims — three.js ### Haber process (three.js) <div class="microsim-player"> <iframe src="https://wikitube-3d-microsims.netlify.app/Haber_process.html" width="100%" height="620" frameborder="0" loading="lazy" sandbox="allow-scripts allow-same-origin" title="Haber process — three.js microsim"></iframe> </div> **Open it full-screen:** [Haber_process.html](https://wikitube-3d-microsims.netlify.app/Haber_process.html) · library `threejs` · route `microsim/threejs/` ### Related microsims Live sims on neighbouring articles: - [[Hydronium]] - [[Electrolysis_of_water]] - [[Fuel_cell]] - [[Interstellar_medium]] *Sim hosted off-article; the article owns the reference, not the runtime (WIKI_RULES §10.4). Placed by `g08_place_microsims.py`.*  ## Overview Nitrogen is seventy-eight per cent of the air and almost nothing alive can use it. N2 is held shut by a triple bond of 945 kJ/mol, the strongest in any common gas, and that inertness is why fixed nitrogen limited world agriculture until the twentieth century. The Haber process opens it with hydrogen over iron, making [[Ammonia]]. It is built from a contradiction worth stating precisely. N2 + 3H2 -> 2NH3 is exothermic -- dH = -92.2 kJ/mol as written, at 298 K -- and turns four molecules of gas into two, so Le Chatelier wants low temperature and high pressure. But where the equilibrium is generous nothing happens: the reaction is thermodynamically allowed and kinetically dead, because something must pay 945 kJ/mol. Every plant therefore runs hot, at 400 to 500 C, throwing away most of the available yield to buy a rate. The process is the resolution of that conflict, and what resolves it is a surface. ## The physics Working pressure is 150 to 300 bar. Even at 200 bar the exit gas carries only about 15 per cent ammonia -- the classic single-pass figure -- so it is condensed out and the unreacted gas recompressed and returned. The loop, not the pass, achieves high conversion. The catalyst is iron, fused from magnetite with promoters and reduced in the converter, and its activity belongs not to "iron" but to particular faces: Fe(111) : Fe(100) : Fe(110) = 418 : 25 : 1 (Spencer, Schoonmaker and Somorjai, *J. Catal.* **74**, 129 (1982)), more than two orders of magnitude between faces of one metal. The reason is geometric: Fe(111) is rough, exposing three atomic levels at once, and carries **C7 sites** -- iron atoms with seven nearest neighbours, down in troughs where an arriving N2 touches several at once and lets them share the bond energy. Fe(110) is close-packed and has none. Two promoters do two jobs, and conflating them is a common error. **K2O is electronic**: potassium donates charge to the iron, which back-donates into the antibonding orbital of adsorbed N2. **Al2O3 is structural**: it keeps the finely divided iron from sintering at 500 C. One raises the rate per site, the other keeps the sites. The mechanism was established by Gerhard Ertl, who took the 2007 Nobel Prize in Chemistry for studies of chemical processes on solid surfaces. It is a Langmuir-Hinshelwood cycle -- everything reacts adsorbed: N2(g) + 2* -> 2 N* dissociative, and rate-limiting H2(g) + 2* -> 2 H* fast, essentially unactivated N* + H* -> NH* + * NH* + H* -> NH2* + * NH2*+ H* -> NH3* + * NH3* -> NH3(g) + * The rate-limiting step is *dissociative adsorption of N2*: the triple bond comes apart on the surface into two separate adsorbed atoms before any hydrogen touches it, and all that follows is stepwise hydrogenation of an atom. N* is hard to make and hard to remove -- H* can leave again as H2, N* must be walked off through NH*, NH2* and NH3* -- so it dominates the working surface, whence the rate law's inverse ammonia dependence. The credit runs backwards from the chemistry. Haber made ammonia from its elements over osmium at Karlsruhe in 1909 and took the 1918 Nobel Prize; Mittasch at BASF screened twenty thousand samples to reach promoted iron; Bosch, who solved the steel problem and built the loop, shared the 1931 prize. Ertl explained why it works, ninety-eight years on. A second error is near-universal in school treatments: the catalyst does not shift the equilibrium, only how fast it is approached, lowering forward and reverse barriers equally. Promoted iron has an apparent activation energy near 100 kJ/mol against 230 to 420 for the gas-phase path: a rate ratio of 1e-16 at 450 C. That number is catalysis. The scale makes the chemistry a geopolitical fact. Production is roughly 180 Mt a year; synthesis takes about 2 per cent of *total final energy consumption* worldwide, 8.6 EJ, and emits some 450 Mt of CO2 directly plus 170 Mt indirectly -- 2.4 t per tonne of ammonia. About 70 per cent goes to fertiliser, the rest to explosives, plastics, mining chemicals and refrigerant. The claim that roughly half of humanity is fed by nitrogen fixed this way is an *estimate* with a lineage, not a measurement: Smil's nitrogen balance in *Enriching the Earth* (2001) gave about 40 per cent for 2000, Erisman *et al.* (*Nature Geoscience* **1**, 636 (2008)) about 48 per cent for 2008, later work about 50 per cent for 2019 -- roughly four billion people, on those authors' accounting. Almost all that CO2 comes not from the synthesis but from the hydrogen, whose feedstock today is overwhelmingly steam methane reforming: strip hydrogen from natural gas, vent the carbon. Ammonia is the largest single consumer of dedicated hydrogen, which is why "green ammonia" changes the feedstock and not the process -- the converter runs unaltered -- and rests entirely on [[Electrolysis_of_water]]. ## Controls -> what each maps to | Control | Maps to | Range / values | Physical meaning | |---|---|---|---| | Face | exposed crystal face | Fe(111) / Fe(100) / Fe(110) | Rebuilds the slab, recounts C7 sites from geometry, sets activity on the 418 : 25 : 1 scale | | T | temperature | 250 to 700 C, default 450 | The master control and the argument itself: yield falls, rate climbs | | P | total pressure | 1 to 400 bar, default 200 | Four moles of gas become two, so pressure buys yield outright | | K2O | electronic promoter | on / off | A per-site rate factor of 2; it changes site chemistry, so it shows on the slab | | Al2O3 | structural promoter | on / off | A site-count factor of 8; it moves the bulk rate, leaving the slab alone | | catalyst | Ea | on / off | Switches activation from ~100 kJ/mol to 320; at 450 C the gas rate is 1e-16 of it | | rate/yield panel | -- | on / off | Falling yield, climbing rate, their humped product | | gas | -- | on / off | Hides the gas phase to clear the view of the slab | | Speed | screen event rate | log slider, 0.01x to 100x | Cosmetic: the surface time base is arbitrary, in no real seconds | | running | -- | on / off | Pauses the kinetic Monte Carlo on the slab | The HUD's headline live quantity is single-pass productivity -- the ammonia mole fraction leaving a bed of fixed size, with the temperature that maximises it alongside. Hunting that maximum is the exercise, because it *is* the process. ## Learning objective After playing, a learner can state the thermodynamic conflict in both directions, explain why a plant deliberately runs at a temperature that lowers its yield, name dissociative N2 adsorption as rate-limiting and N* as the dominant surface species, and tell an electronic promoter from a structural one. ## Limits and connections The thermodynamics is computed, not quoted: NIST Shomate polynomials give dG(T) and ln K, and the ammonia fraction comes from bisecting f^2/(1-f)^4 = (27/256) K P^2 on every slider move. The sim takes dfH(NH3) = -45.898 kJ/mol for internal consistency, so its printed dH at 298 K is about -91.8, not the textbook -92.2 that comes from the more commonly quoted -46.11. And the ideal gas fails at pressure: against Larson and Dodge's 1923-24 tables it holds to 5 per cent at 100 atm but runs 20 per cent low at 300, so one fitted fugacity term brings every entry within 8 per cent. It is the only fitted quantity here. The converter is a plug-flow bed integrated live with the Temkin-Pyzhev rate equation at alpha = 0.5, fed 3:1 gas with 3 per cent recycled ammonia; real loop gas also carries 10 to 15 per cent inerts, and omitting them flatters the yield. Multiplying rate by yield and hunting a maximum does not work -- that needs an activation energy under about 55 kJ/mol against a real 100 -- so the optimum comes, as in a real converter, from the reverse reaction in a finite bed. One constant, bed size times activity, is set so promoted Fe(111) at 450 C and 200 bar delivers the classic 15 per cent; everything else falls out, the optimum landing at 490 C and moving to 570 and 665 C on the other two faces. The surface is the honest weak point. Events on the slab are a lattice kinetic Monte Carlo on the real site geometry, one rate constant per elementary step, with adsorption scaled by partial pressure and face activity. The *dependences* are right -- hotter runs faster, pressure fills the surface, Fe(110) starves for nitrogen and ends up covered in hydrogen -- but the absolute time base is not, its prefactors chosen to make the dance watchable. Hence no surface time and no turnover frequency in seconds anywhere in the HUD; the quantitative claims live in the thermodynamics and reactor model. The hydrogen this consumes is why [[Electrolysis_of_water]] and [[Fuel_cell]] sit on the same hub. ## Wikipedia : Wikitube **Strict pair:** [Wikipedia](https://en.wikipedia.org/wiki/Haber_process) : [Wikitube](https://en.wikitube.io/wiki/Haber_process) ## Previous hub tags Tree parent: [[Hydrogen]]. Legacy hubs: `HYDROGEN`. --- *Created 2026-08-05 - append-only - authored to WIKI_REPOPULATION_PROTOCOL v1.0 section 5 - portal-microsim-pass (PORTAL_Hydrogen batch 2) - 0 deletions*