Hydrogen Refuelling Station Adsorbent Selection: SAE J2719 Grade D Compliance at 700 bar

1. What SAE J2719 and ISO 14687 Grade D actually require

SAE J2719 (latest revision 2020-09) and ISO 14687:2019 set the fuel quality spec for proton exchange membrane (PEM) fuel cell vehicles. The two standards are technically aligned on the key numbers: minimum H2 purity 99.97% (molar), total non-hydrogen contaminants below 100 umol/mol, water below 5 umol/mol, oxygen below 5 umol/mol, total sulfur compounds below 0.004 umol/mol. ISO 14687 adds stricter documentation: the station operator must keep a per-batch QC log of the upstream H2 source, the adsorbent change-out dates, and the regenerator cycle counter. The downstream adsorbent train must be designed so that, at end-of-cycle (EoC) bed saturation, the dispenser outlet still meets Grade D. End-of-life is defined at 90% of the dynamic adsorption capacity, not 100% - that 10% headroom is the only safety margin against a regeneration slip.

2. Why 5A molecular sieve is the standard HRS primary dryer choice

5A (LTA zeolite, Ca-exchanged) has a pore aperture of approximately 5 angstroms. That aperture is large enough to admit water (kinetic diameter 2.6 angstroms) and CO2 (3.3 angstroms) but small enough to exclude H2 (2.9 angstroms, but the molecule is rejected by the LTA cage geometry at room temperature). The practical consequence: 5A does not co-adsorb H2, so there is zero risk of adsorbent carryover into the dispenser and zero H2 inventory tied up in the bed. By contrast, 4A sieve will co-adsorb 0.2-0.4% of H2 at 70 MPa, which is both a safety issue (a 5A bed at 70 MPa has measurable H2 inventory in its pores) and an economic issue (that H2 is not recoverable in the regeneration purge).

The 8x12 mesh (2.4-3.4 mm bead) is the industry default for HRS. Smaller beads (10x18) give faster kinetics but cost 0.4 bar more pressure drop at design flow. Larger beads (6x8) save pressure drop but reduce the breakthrough time by 30-40% and force more frequent regeneration. The 700 bar dispenser set has a hard limit of 0.5 bar pressure drop total - so the 8x12 mesh sits right in the middle.

3. Activated alumina as the sulfur and amine guard

Upstream of the 5A bed, a 5-8 kg activated alumina guard removes any sulfur compounds, amines, and high-molecular-weight hydrocarbons that slip from the electrolyzer (if PEM or alkaline) or the SMR + PSA (if grey/blue H2). The activated alumina acts as a sacrificial bed: when it saturates with sulfur (typically 0.1-0.3 wt% breakthrough), the 5A primary bed is still protected. The guard is replaced every 6-12 months; the 5A bed every 3-5 years.

For more on the underlying alumina grades used in industrial drying trains, see our coverage of regeneration gas composition for activated alumina.

4. Bed sizing for a 350 kg/day station

A typical 350 kg/day station has 2 parallel compression trains, each sized for 180 kg/day at design flow. Each train uses 110-140 kg of 5A molecular sieve (8x12 mesh) plus 5-8 kg activated alumina guard, in a vessel 400 mm diameter x 1200 mm tall (L/D 3.0). The breakthrough time at design flow is 8-12 hours, which sets the regeneration cycle. At end-of-cycle, the bed is regenerated with heated N2 purge at 280-300 degC for 2-3 hours, purge:feed ratio 8-12%.

Table 1 - HRS adsorbent train sizing reference (350 kg/day class)

5. Pre-cooling adsorbent for 700 bar service

At 70 MPa, the H2 from the dispenser is pre-cooled to -40 degC per SAE J2600 and ISO 19880-1 to limit the temperature rise during vehicle tank filling (the Joule-Thomson effect on the vehicle-side regulator). The pre-cooler typically includes a small 3A or 4A molecular sieve canister (0.3-0.6 kg of 1.6-2.5 mm bead) upstream of the chiller, regenerated every 200 cycles or 30 days. The duty is tiny but its failure is the most common out-of-spec root cause seen at HRS commissioning audits.

6. Regeneration cycle design

For 5A primary beds, regeneration is heated N2 purge at 280-300 degC for 2-3 hours. The purge:feed ratio of 8-12% is the binding economic constraint - that N2 (and the H2 it strips from the bed) becomes a vent stream that must be either recovered (90% of HRS do this via a back-to-compressor return line) or flared. Skipping regeneration to extend cycle life saves adsorbent but costs more in H2 recovery from the purge stream than the adsorbent savings.

Regeneration scheduling is typically: 8 h adsorption, 2 h heating, 1 h cooling, 1 h standby = 12 h cycle. With 2 parallel trains, one is always on adsorption while the other is on regen, so the station delivers continuous flow.

7. ISO 19880-8:2019 commissioning tests

ISO 19880-8:2019 Section 6 specifies 3 on-site tests at HRS commissioning: (1) H2 purity per SAE J2719 Annex B at the dispenser nozzle, (2) pressure drop test at design flow, (3) leak test of the entire adsorbent train. A 4th test, dehumidifier bed breakthrough, is industry practice: operate at 1.1x design flow for 4 hours and measure outlet dew point - if it rises above -62 degC, the bed is undersized.

8. TCO comparison - molecular sieve vs activated alumina for HRS

For a 350 kg/day station, 5A molecular sieve costs USD 4,800-6,200 per train (140 kg, FOB China), activated alumina USD 800-1,200 for the guard bed. The molecular sieve is 5x more expensive but does 90% of the work. The activated alumina guard is the cheap insurance against amine and sulfur slip from the upstream electrolyzer or SMR. The combined cost is roughly 1.8-2.4 cents per kg of H2 dispensed over 5 years.

Three cost items that station owners consistently underestimate are: (a) the on-site engineer visit from the adsorbent supplier (typically USD 4,500-7,000 per trip including travel and per-diem, for a 2-3 day visit), (b) the spare-bed cost during regeneration, since most stations keep a spare 5A vessel ready to swap in to keep the train on line, and (c) the disposal cost of spent activated alumina guard (classified as non-hazardous industrial waste in the EU, but a regulated waste stream in parts of the US and Japan that can cost USD 200-400 per ton for disposal). The net effect is that the headline purchase price represents only 55-65% of the true 5-year TCO. The other 35-45% is service, spare inventory, and disposal. Specifying the right adsorbent from the start, and using a supplier with documented on-site engineering capability, materially reduces the second and third cost categories.

Table 2 - 5-year TCO per train at 350 kg/day

9. Common failure modes in the field

From our service records on 12 commissioned HRS in Europe and Asia, the top 5 failure modes in the first 18 months of operation are: (1) activated alumina guard exhausted prematurely (sulfur slip from the SMR), (2) 5A bed caked by upstream oil carryover from the compressor, (3) pre-cooler 3A bed plugged by carryover from the dryer, (4) vessel corrosion under the activated alumina bed (chloride pitting from the H2O2 carryover of the electrolyzer), (5) regeneration N2 heater undersized, leading to incomplete regen and early breakthrough. Items 1, 4 and 5 each account for ~20% of the failure budget. For sulfur poisoning recovery, our related post on H2S poisoning detection and recovery covers the diagnostic sequence.

10. What station EPC contractors should ask the adsorbent supplier

8 questions we recommend station EPCs put in the RFQ: (1) what is the lot-certified dynamic H2O capacity at 35 degC, 7 bar? (2) what is the crush strength per ASTM D4179 (target > 5 daN per bead for 8x12)? (3) what is the LOI at 1000 degC (target < 1.5 wt%)? (4) does the bead pass ISO 9001 traceability from raw powder to final packing? (5) is the activated alumina guard shipped in vacuum-sealed foil to prevent pre-adsorption? (6) what is the recommended regeneration profile? (7) is the manufacturer willing to support on-site commissioning? (8) does the price include technical service for the first 12 months?

Table 3 - RFQ acceptance criteria for 5A molecular sieve at HRS

11. Procurement and supply notes

11.1 Why adsorbent lot consistency matters more than price for HRS

Fuel cell vehicle OEMs (Toyota, Hyundai, Honda, Daimler) reject a station's H2 on the basis of a single failed analyzer reading - which means the entire upstream adsorbent train is at risk of being blamed. In practice the most expensive failure mode for an HRS operator is not the cost of the sieve or alumina, it is the cost of a 14-day shutdown while the supplier dispatches a replacement batch, the EPC contractor flies a commissioning engineer back on site, and the OEM audits the documentation chain. A station that has a 2-3 year track record with consistent adsorbent lot performance can price its dispensed H2 at a small premium because fleet operators trust the uptime. The corollary: a low-bid adsorbent from an unknown supplier, even with a passing certificate, is a real economic risk.

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 H2O capacity (which is typically 20-22 wt% and stays within +/- 0.5 wt% batch to batch) but the attrition rate, which can swing from 0.2 wt% to 1.1 wt% in poorly controlled production. A 0.2 wt% attrition increase is invisible at lab scale but translates to several kilograms of dust per year accumulating at the bed support screen - which in turn causes high pressure drop alarms 6 months earlier than design. The reverse engineering is simple: if your station's pressure drop is climbing 50% faster than design, suspect attrition before you suspect coking.

Table 4 - Lot acceptance test panel for 5A molecular sieve (HRS class)

11.2 Storage and shelf-life notes

5A molecular sieve and activated alumina are both hygroscopic. From the moment they leave the factory dryer, they start picking up moisture from ambient air. A typical 5A bead will pick up 1.5-2.5 wt% moisture per hour of open exposure at 25 degC / 70% RH. At the factory, the sieve 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 dynamic capacity is reduced by 8-12%.

Best practice at the HRS site: bring the unopened drum into the compressor room 24 hours before the planned change-out, allow it to warm to room temperature (this prevents condensation on the cold beads when the drum is opened), and load the vessel within 4 hours of opening. If loading is delayed beyond 4 hours, the drum should be re-sealed with fresh desiccant and the bed capacity re-rated. Loading crews that have not worked with sieve before sometimes leave drums open overnight, which can drop the effective capacity by 30% in humid climates. For a deeper dive into adsorbent storage mistakes, our related post on 7 common storage and handling mistakes catalogues the rest of the typical issues.

AluminaWorld supplies 5A molecular sieve (8x12 mesh, 4x6 mesh, 10x18 mesh) from a dedicated line at our Zibo facility, with FOB / CIF / DDP terms to the main HRS EPC hubs in Germany, South Korea, Japan, and the US Gulf Coast. Standard lead time is 18-22 days for a 350 kg/day class order (4-6 MT of 5A + 200-400 kg of activated alumina guard). We support ISO 19880-8 commissioning audits with a 2-day on-site engineer at no charge for orders above 3 MT. Free samples of 200 g to 2 kg are available on request.

12. Next steps

If you are specifying an HRS adsorbent train for a 350-700 kg/day class station, or auditing an existing supplier for renewal, contact our team for a tailored bed sizing calculation and a sample shipment. We can also provide reference documentation for SAE J2719 / ISO 14687 / ISO 19880-8 compliance review.