Activated Alumina Ammonia Cracking Hydrogen Carrier — What It Actually Is (And What It Is Not)

Abstract. Ammonia (NH3) is one of the most attractive hydrogen carriers for long-distance and seasonal storage because it is CO2-free at point of use, already traded globally at scale, and cracks back to 75% H2 / 25% N2 at 650-850 degC over a nickel or ruthenium catalyst. The problem downstream is residual NH3: a well-operated catalytic cracker leaves 50-200 ppmv of NH3 in the product stream, against a fuel cell spec of 0.1 ppmv (ISO 14687:2019 Grade D / SAE J2719-202009). This article walks through the activated alumina polishing bed that bridges that 3 to 4 orders-of-magnitude gap - bed sizing for a 400 kg/day cracker, the rationale for a layered 5A + activated alumina train, regeneration duty at 280 degC with N2 purge, FTIR/TDLAS sampling, and 8 RFQ questions EPC contractors should put to the adsorbent supplier. Activated alumina , 5A molecular sieve , and 13X molecular sieve are the three grades that come up most often in NH3 polishing design. For H2 refuelling station context, see our related post on SAE J2719 Grade D compliance at 700 bar ; for electrolyzer dew-point context, see molecular sieve drying for green H2 electrolyzers .

Ammonia has carried the torch as a hydrogen carrier in the fertilizer and chemicals industry for over a century, with global trade in 2024 around 18 Mt NH3 and a maritime shipping fleet dedicated to it. The case for NH3 as a low-carbon hydrogen carrier rests on three numbers: 17.6 wt% H2 content by mass, a liquefaction temperature of -33 degC at ambient pressure (vs -253 degC for liquid H2), and a global trade infrastructure already in place. Cracking NH3 back to H2 + N2 is the reverse of the Haber-Bosch synthesis, and the catalyst technology - nickel or ruthenium on alumina - is mature. The remaining engineering question is polishing: removing the residual NH3 down to fuel-cell grade.

Three numbers frame the polishing bed duty. A well-operated catalytic cracker at 650-750 degC, 5-30 bar, over a nickel-on-alumina catalyst, achieves 99.5-99.8% NH3 conversion. The 0.2-0.5% that does not convert leaves 50-200 ppmv NH3 in the product stream at the cracker outlet, depending on feed purity, space velocity, and catalyst age. ISO 14687:2019 Grade D, mirrored by SAE J2719-202009, sets the maximum NH3 contaminant at 0.1 umol/mol = 0.1 ppmv. The polishing bed must therefore remove 99.5% to 99.95% of the residual NH3 in a single pass - which is beyond what a single 5A molecular sieve bed can do at reasonable regeneration duty.

Activated Alumina Ammonia Cracking Hydrogen Carrier — The Hidden Defect / Failure Mode

5A (LTA zeolite, Ca-exchanged) has a pore aperture of ~5 angstroms, large enough to admit NH3 (kinetic diameter 2.6 angstroms) into its cage structure. Its equilibrium NH3 capacity at 250 degC is roughly 6-8 wt% - higher than activated alumina on a per-mass basis. The problem is regeneration: NH3 binds to the acid sites of 5A with an enthalpy of ~70 kJ/mol, which is at the upper edge of what a 280-300 degC N2 purge can reverse economically. Stripping NH3 fully from a 5A bed requires 350-400 degC regen, which raises energy cost, pushes the heater metallurgy into stainless, and shortens adsorbent life by 30-40%.

The practical workaround is a layered bed: 5A upstream to do the bulk of the NH3 removal (~80%), activated alumina downstream to polish the last 0.5-2 ppmv down to below 0.1 ppmv. The activated alumina's NH3 bond is ~40 kJ/mol, weaker because the Al2O3 surface has fewer strong acid sites than the zeolite, and that lower bond energy means a 250-280 degC N2 purge can fully strip the bed. The two beds regenerate on independent cycles, so the polishing train can run continuously even when one bed is on regen.

• The problem downstream is residual NH3: a well-operated catalytic cracker leaves 50-200 ppmv of NH3 in the product stream, against a fuel cell spec of 0.1 ppmv (ISO 14687:2019 Grade D / SAE J2719-202009).

Activated Alumina Ammonia Cracking Hydrogen Carrier — Measured Numbers Across Grades

Three alumina grades are used in NH3 polishing service. AA300 (a 3-5 mm sphere with BET ~300 m2/g, pore volume ~0.45 cm3/g, and bulk density 750-820 kg/m3) is the industry default for cracker polishing at 250-300 degC. AA350 (a 5-7 mm sphere with BET ~350 m2/g, higher pore volume, bulk density 720-780 kg/m3) gives 15-20% more NH3 capacity but at higher pressure drop. AA700 (a 1-2.5 mm bead with BET ~360 m2/g, used in liquid-phase polishing columns) is too small for gas-phase cracker duty at typical space velocities - it creates excessive pressure drop and dust generation.

The pellet chemistry matters. Standard gamma-alumina (gamma-Al2O3) is the workhorse. For high-temperature service above 350 degC, alpha-alumina-stabilized grades (Al2O3 doped with 3-5 wt% SiO2 or La2O3) keep their surface area through more regeneration cycles. The surface acid site density - measured by NH3-TPD (temperature-programmed desorption) - should be in the 0.15-0.30 mmol/g range for cracker polishing. Higher acid site density means higher NH3 capacity but also stronger binding, which pushes regeneration duty up.

Where Activated Alumina Ammonia Cracking Hydrogen Carrier Earns Its Premium — Core Chemistry

A typical 400 kg/day H2 cracker at 650 degC, 5-8 bar outlet, with feed NH3 of 99.5% purity and cracker conversion of 99.6%, has a residual NH3 slip of 100-150 ppmv at the cracker outlet before polishing. The polishing bed train is sized to reduce this to below 0.1 ppmv at the H2 product outlet. With a layered 5A + activated alumina train:

Upstream 5A bed: 80-120 kg of 5A (8x12 mesh, bead) in a 350 mm diameter x 1100 mm tall vessel (L/D 3.1). This bed removes the bulk of the residual NH3 to ~2-5 ppmv. Adsorption cycle: 4 hours at design flow. Regeneration: heated N2 purge at 320 degC for 3 hours, purge:feed ratio 12-15%.

Downstream activated alumina bed: 60-90 kg of AA300 (3-5 mm sphere) in a 350 mm diameter x 900 mm tall vessel (L/D 2.6). This bed polishes the 5A outlet from 2-5 ppmv to below 0.1 ppmv. Adsorption cycle: 4 hours at design flow. Regeneration: heated N2 purge at 280 degC for 3 hours, purge:feed ratio 15-20%.

Activated Alumina Ammonia Cracking Hydrogen Carrier — Dosing Across Service Conditions

Cracker operating pressure has a first-order effect on polishing bed sizing because the partial pressure of residual NH3 is set by total pressure times mole fraction. A 20 bar cracker outlet sees roughly 4x the NH3 partial pressure of a 5 bar outlet, which means the polishing bed must hold roughly 4x the NH3 mass per cycle. In practice, modern crackers for downstream compression run at 5-8 bar outlet, balancing compression cost downstream vs polishing bed sizing upstream. Older Haber-Bosch-decommissioned ammonia plants being retrofitted as crackers often operate at 30 bar because they reuse the existing synthesis loop, but those are typically supplying industrial H2 users (not fuel cell grade) and can tolerate 1-5 ppmv NH3 slip.

Two consequences of running the cracker at higher pressure: first, the polishing bed L/D ratio must increase from 2.6-3.1 to 3.5-4.0 to maintain the breakthrough time at design flow. Second, the regeneration duty scales roughly linearly with the operating pressure because the purge gas requirement is set by the moles of NH3 to be stripped, not by the moles of carrier gas. For a 30 bar cracker train, the regen duty at 280 degC is approximately 0.9-1.4 kWh per kg of activated alumina per cycle.

Hydration Kinetics, Set Control, and Working Time

Energy consumption is 0.8-1.2 kWh per kg of activated alumina per regen cycle. For a 60-90 kg bed, that is 50-110 kWh per regen cycle, or roughly 4-8 MWh per train per year at continuous duty. At industrial electricity prices of USD 0.08-0.12 per kWh, the regeneration energy cost is USD 4,800-6,400 per train per year. This is roughly 70-80% of the polishing bed's total annual operating cost.

Three cost items that crackers consistently underestimate are: (a) the spare activated alumina drum during regeneration, since most operators keep a spare 60-90 kg drum ready to swap into a second vessel to keep the train on line, (b) the FTIR analyzer maintenance - the sample line and analyzer lamp need replacement every 12-18 months at USD 800-1,400 per service, and (c) the N2 supply - if the cracker does not have a dedicated liquid N2 tank, the regen duty is being served by pipeline N2 at higher cost. The net effect is that the polishing bed represents 1.5-2.5 cents per kg of H2 produced, on top of the cracker operating cost itself.

1. Step 1: Energy consumption is 0.8-1.2 kWh per kg of activated alumina per regen cycle. For a 60-90 kg bed, that is 50-110 kWh per regen cycle, or roughly 4-8 MWh per train per year at cont...
2. Step 2: Three cost items that crackers consistently underestimate are: (a) the spare activated alumina drum during regeneration, since most operators keep a spare 60-90 kg drum ready to sw...
3. Step 3: Measuring 0.1 ppmv NH3 in a hydrogen-rich stream is not trivial. Three analytical techniques are in industrial use. FTIR (Fourier Transform Infrared Spectroscopy) is the workhorse ...

• Ammonia has carried the torch as a hydrogen carrier in the fertilizer and chemicals industry for over a century, with global trade in 2024 around 18 Mt NH3 and a maritime shipping fleet dedicated to it.
• Cracking NH3 back to H2 + N2 is the reverse of the Haber-Bosch synthesis, and the catalyst technology - nickel or ruthenium on alumina - is mature.
• The remaining engineering question is polishing: removing the residual NH3 down to fuel-cell grade.

Procurement Specification — What to Demand on the COA

The sample line is the most common source of false-low readings. NH3 adsorbs on cold stainless steel surfaces (below 80 degC), biasing the analyzer reading 30-60% low. The fix is a heated sample line maintained at 150 degC throughout, with the shortest possible run from the bed outlet to the analyzer (ideally under 3 m). The sample pump must be downstream of the analyzer cell, not upstream, so the cell operates at low pressure to maximize residence time. A bypass loop with a fast-acting needle valve is essential for zeroing the analyzer against N2 every 4-6 hours.

At commissioning, the polishing bed must demonstrate outlet NH3 below 0.05 ppmv for 24 hours continuous at 1.1x design flow. This is the binding test in ISO 14687:2019 Grade D documentation for any cracker skid destined for fuel cell application. Failure to meet this test means the bed is undersized - typically a 30-50% increase in bed mass or a switch to AA350 grade activated alumina is required to recover.

• Three numbers frame the polishing bed duty.
• A well-operated catalytic cracker at 650-750 degC, 5-30 bar, over a nickel-on-alumina catalyst, achieves 99.5-99.8% NH3 conversion.
• The 0.2-0.5% that does not convert leaves 50-200 ppmv NH3 in the product stream at the cracker outlet, depending on feed purity, space velocity, and catalyst age.
• ISO 14687:2019 Grade D, mirrored by SAE J2719-202009, sets the maximum NH3 contaminant at 0.1 umol/mol = 0.1 ppmv.

Storage, Shelf Life, and Inventory Hygiene

The fix sequence for item 1 (the most common) is to first confirm that the cracker catalyst is at its end-of-life conversion (typically 99.2-99.4% after 18-24 months). If yes, swap the catalyst and re-baseline the polishing bed. If no (catalyst still showing 99.6%+ conversion), the polishing bed is undersized and must be expanded by 30-50%. For items 2 and 5, the corrective action is to add or replace the upstream 5 um sintered metal particulate filter and verify the cracker feed NH3 purity is 99.5%+ with no chloride content above 0.5 ppmw. For more on these patterns in the parallel electrolyzer polishing context, our related post on molecular sieve drying for green H2 electrolyzers covers a similar diagnosis sequence.

8 questions we recommend cracker EPCs put in the RFQ: (1) what is the lot-certified NH3 capacity at 250 degC, 5 bar, 100 ppmv inlet NH3 (target > 1.5 wt% on activated alumina)? (2) what is the crush strength per ASTM D4179 (target > 6 daN per bead for 3-5 mm sphere)? (3) what is the regeneration cycle endurance - how many adsorption/regen cycles does the bed tolerate before NH3 capacity drops below 80% of initial (target > 5,000 cycles)? (4) is the supplied alumina in the gamma or alpha phase, and what is the BET surface area (target 280-330 m2/g for AA300 grade)? (5) does the bead pass ISO 9001 traceability from raw powder to final packing, with a per-lot NH3-TPD curve available? (6) is the activated alumina shipped in vacuum-sealed foil to prevent pre-adsorption? (7) what is the recommended regeneration profile (target 280 degC, 3 h soak, 15-20% purge:feed)? (8) is the manufacturer willing to support on-site commissioning and FTIR cross-calibration with the analyzer contractor?

• Moisture pickup in unsealed drums above 60% RH warehouse humidity (capacity loss in 30-90 days).
• Cross-contamination between activated alumina and molecular sieve stored in shared warehouses.
• Thermal-shock spalling when cold drums (-10 C winter) are opened into warm air.

• Store in sealed drums or foil-lined supersacks; re-seal immediately after partial withdrawal.
• Keep warehouse RH below 50% — climate control if you are in a tropical zone.
• If a drum has been open more than 60 days, re-test moisture and capacity before deploying into service.
• For bulk shipments above 5 t, request sealed tanker delivery with inline desiccant vent.

Total Cost of Ownership — Why Premium Pays for Itself

The practical check is to ask the supplier for the last 6 months of batch certificates, cross-checked against the bed change-out dates. The variability to watch is not the headline NH3 capacity (which is typically 1.6-1.9 wt% and stays within +/- 0.1 wt% batch to batch) but the NH3-TPD acid site density, which can swing from 0.12 mmol/g to 0.38 mmol/g in poorly controlled production. A 0.12 mmol/g acid site density reading is invisible at lab scale but translates to a 30-40% drop in service NH3 capacity at the cracker outlet, which in turn causes early breakthrough and out-of-spec H2 within the first 60-90 days of operation. The reverse engineering is simple: if your FTIR reading is climbing earlier than design, suspect acid site density before you suspect the FTIR itself.

Activated alumina and 5A molecular sieve are both hygroscopic. From the moment they leave the factory dryer, they start picking up moisture from ambient air. A typical AA300 bead will pick up 1.2-2.0 wt% moisture per hour of open exposure at 25 degC / 70% RH. At the factory, the alumina is sealed in plastic-lined steel drums or aluminum-laminate bags with a desiccant sachet and a vacuum pull. Once on site, the unopened drum has a shelf life of 12 months - after that, the bead has picked up enough moisture that the in-service NH3 capacity is reduced by 8-12%.

The math: a 5-10% longer service life plus lower thermal-cycling spalling plus reduced replacement-frequency combine to deliver 12-25% lower TCO over a single cell-stack replacement cycle, even before the safety/insurance argument for eliminating thermal-shock microcracks. The 8-15% raw-material premium pays back 1.5-2.5x over the first cycle.

Related Products & Technical Resources

We supply activated alumina, molecular sieve, and alumina-ceramic products as part of a broader specialty adsorbent catalog. The full datasheet for this grade includes additional grain-size distributions, customer-specific blend recommendations, and pilot-plant support. Talk to our technical team if your application has unusual thermal-cycling, mechanical-load, or chemical-feed constraints.

• Activated Alumina — gamma-Al2O3 adsorbent for CO2, H2S, and fluoride removal.
• Molecular Sieve — 3A / 4A / 5A / 13X grades for PSA, TSA, and drying service.
• Alumina Powder — high-purity Al2O3 powders for general industrial use.
• Aluminum Hydroxide — ATH flame retardant grade.
• Reactive Alumina Low-Soda for Castable and Burned-Magnesia Bonding
• MCFC CO2 Removal: Activated Alumina vs Carbon Molecular Sieve
• Adsorbent Bed Pressure Drop: Ergun Equation Design Guide

Next Steps — Get a Sample or Talk to Our Engineers

For samples, COAs, or application-specific blend recommendations, reach out to the Aluminaworld technical team. We hold stock of the relevant grade in our Zibo warehouse with same-week dispatch for trial orders, and we can also produce custom recipes for OEMs and EPCs with multi-ton annual demand.

Contact: WhatsApp +86 13325222240 (preferred for fast reply) or email barry@aluminaworld.com. Trial orders ship FOB Qingdao in 7-10 days; full-container shipments in 15-25 days depending on grade and customization.