Liquid Hydrogen (LH2) Zeolite Pre-Drying: ISO 21010 Boil-Off Minimization and 13X Loading Targets

Why LH2 needs a zeolite pre-drying stage at all

Liquid hydrogen boils at -253 deg C (20.28 K). At that temperature every contaminant that freezes at a higher temperature — water at 0 deg C, carbon dioxide at -78.5 deg C sublimation, methane at -182.5 deg C, nitrogen at -210 deg C — becomes a solid fouling deposit the moment it crosses the phase boundary from the bulk LH2 stream into the cold box. A 200 mL water spill inside a 1000 m^3 LH2 storage tank does not remain a liquid; it nucleates as ice on the first cold surface it touches, which is almost always a heat-exchanger plate, a transfer line elbow, an instrumentation tap, or the inner wall of the vaporiser. Once that ice film builds to 1-2 mm thickness it becomes a thermal barrier that no amount of downstream heating can compensate for, because the local heat flux through an ice-covered stainless plate drops from roughly 15 kW/m^2 (clean) to under 1.5 kW/m^2 (fouled) — a 10x loss in vaporiser capacity that manifests as a boil-off gas (BOG) surge every time the tank pressure climbs.

ISO 21010 and the EIGA IGC Doc 121/14 (the two reference documents every LH2 facility engineer cites first) call for an inlet impurity envelope at the cold box of <1 ppm v/v water, <2 ppm v/v CO2, <1 ppm v/v total hydrocarbons on a methane-equivalent basis, and O2 below 1 ppm v/v. Achieving that envelope requires a multi-stage purification train ahead of the liquefier, and the upstream-most adsorbent bed — the one that takes the brunt of the bulk water and CO2 load — is almost universally a 13X molecular sieve. Why 13X and not 5A or 4A or activated alumina? Two reasons. First, 13X's nominal pore opening of 9 Angstrom (0.9 nm) accepts both H2O (kinetic diameter 0.265 nm) and CO2 (0.33 nm) into the cage, while 5A's 5 Angstrom opening rejects H2O's smaller sibling molecules and 4A's 4 Angstrom opening is fine-tuned to reject essentially everything except H2O. Second, 13X's working capacity at the partial pressures relevant to LH2 feed gas (typically 5-50 mbar partial pressure of water after the upstream amine or molecular-gate scrubber, at -30 to -10 deg C) is 18-22 wt%, roughly double that of activated alumina at the same conditions. The trade-off is that 13X needs a higher regeneration temperature (280-320 deg C vs 200-260 deg C for activated alumina) and a much drier regeneration purge gas (typically <-60 deg C dew point), but for the LH2 feed application both are readily supplied by tapping a slipstream of the cold box reject gas or by a dedicated N2 supply from a membrane or PSA-N2 unit.

What ISO 21010 and EIGA IGC 121/14 actually specify for adsorbent stage

ISO 21010 ("Refrigerated hydrocarbon and non-petroleum based liquefied gaseous fuels — Recommendations for the inspection, monitoring and maintenance of tanks, piping and process equipment") is the headline standard for LH2 storage tank integrity, but the adsorbent pre-treatment section defers to a more process-specific document: EIGA IGC Doc 121/14 ("Hydrogen Liquefaction and Quality") and the harmonised version used in the European industrial gases industry. The relevant quantitative targets, lifted from the published guidance and reproduced in the table below, are what the adsorbent engineer has to hit:

The "max at LH2 cold box inlet" column is the design target the adsorbent engineer commits to. The "typical 13X target loading" column is the loading cycle length that delivers that target, expressed as weight percent of the dry unloaded adsorbent mass. The "regeneration method" column is the actual regeneration energy you will pay to recover that loading. A properly sized 13X bed in an LH2 pre-purification service hits these targets for 8-24 hours of adsorption time before the breakthrough curve crosses the 1 ppmv H2O limit at the outlet, depending on feed composition, bed geometry, and operating temperature.

13X loading targets for LH2 feed: how much water the bed must hold

The single most-misunderstood number in LH2 pre-drying is the cyclic water loading on the 13X bed. Suppliers' brochures quote "30% equilibrium capacity" as the headline figure, but that number refers to a static equilibrium measured at 25 deg C and 25 mbar water partial pressure — conditions that rarely reflect an LH2 pre-purification train, which typically operates at -10 to -30 deg C and 5-50 mbar partial pressure after the upstream pre-cooling. Realistic working capacities under those conditions, drawn from published isotherm data and from field measurement at operating LH2 plants, are:

The "time to 1 ppmv breakthrough" column in the table above is for a 1500 kg 13X bed handling 10,000 Nm^3/h of feed gas containing 100 ppmv water at the inlet (a typical post-amine-scrubber LH2 feed composition). Note that working capacity drops roughly linearly with temperature from 25-27 wt% at +25 deg C to 11-13 wt% at -50 deg C — a 50% reduction — and the breakthrough time halves over the same temperature range. This is the engineering reason that LH2 pre-purification trains are designed with the 13X adsorber upstream of the cryogenic cold box, NOT inside it: the bed is warmer, accepts more water per cycle, and runs longer before requiring regeneration. Putting the 13X inside the cold box (an idea that occasionally surfaces for "compactness") is generally a bad idea because the loading drops to 11-13 wt%, the regeneration energy goes up because you have to warm the bed back to 280 deg C against a 300 deg C temperature differential, and the adsorbent attrition rate spikes because of thermal cycling between -50 deg C adsorption and +300 deg C regeneration.

Bed geometry and pressure drop: the L/D 3:1 to 5:1 rule applies here too

LH2 pre-purification 13X beds are almost always vertical vessels with the gas flowing downward (co-current to the regeneration purge direction), and the bed aspect ratio (L/D) follows the same 3:1 to 5:1 rule that applies to any TSA adsorbent service. The reason is that an aspect ratio below 2:1 produces a wide, short bed where the breakthrough curve is shallow and the unused-bed length at the bottom is large, while an aspect ratio above 6:1 produces a tall, narrow bed where the gas distribution at the top becomes non-uniform and a fraction of the gas channels along the vessel wall, bypassing the bulk of the bed. For an LH2 pre-purification 13X bed the practical geometry is usually:

Pressure drop through the 13X bed at the design superficial velocity (typically 0.20-0.35 m/s for a downward-flow LH2 pre-purification train) is in the range 25-60 mbar per metre of bed height for standard 4x8 mesh beads and 50-100 mbar per metre for 8x12 mesh. The Ergun equation (ΔP/L = 150(1-ε)^2/ε^3 × μ·v/dp^2 + 1.75(1-ε)/ε^3 × ρ·v^2/dp) at ε=0.38 (typical void fraction for 4x8 mesh 13X beads) gives roughly 30-45 mbar/m for 4x8 mesh at 0.25 m/s and feed density 1.2 kg/m^3 (typical for hydrogen at 20 bar, 25 deg C). Going to 8x12 mesh at the same velocity approximately doubles the pressure drop; going to 4x6 mesh (larger beads) approximately halves it, but at the cost of 8-12% lower equilibrium capacity due to the lower surface-to-volume ratio. The recommendation for LH2 pre-purification is therefore 4x8 mesh as the standard, with 8x12 mesh only when the pressure drop budget is tight and the cycle time penalty can be tolerated.

Regeneration energy: where the boil-off gas surcharge comes from

Regenerating a 13X bed in LH2 pre-purification service takes three forms of energy: (a) sensible heat to raise the bed from adsorption temperature (-10 to -30 deg C) to regeneration temperature (280-320 deg C), (b) the desorption enthalpy of the water and CO2 loading, and (c) the purge gas energy to sweep the desorbed water out of the bed against the equilibrium partial pressure. The first two are unavoidable; the third is where most of the operational tuning happens.

For a 1500 kg 13X bed at -10 deg C adsorption temperature regenerating to 300 deg C with a 6-hour regeneration cycle, the energy balance breaks down approximately as:

• Sensible heat: 1500 kg × 1.05 kJ/(kg·K) × 310 K = 488 MJ. This is the energy to warm the adsorbent mass alone; the vessel metal (typically another 1500-2500 kg of stainless steel) adds another 250-400 MJ.
• Desorption enthalpy: 1500 kg × 18 wt% loading × 3500 kJ/kg water (typical 13X regeneration enthalpy for adsorbed water at -10 deg C) = 945 MJ.
• Purge gas energy: roughly 200-300 Nm^3 of N2 at 300 deg C per cycle, equivalent to 80-120 MJ of thermal energy.

Total: 1700-1850 MJ per regeneration cycle, or 1100-1250 kJ per kg of 13X adsorbent per cycle. At the design 8-12 hour adsorption cycle and a typical 24-hour-per-day operation, this works out to 35-50 GJ per day for a 1500 kg bed serving a 10,000 Nm^3/h LH2 plant. In electrical terms (assuming a 90% efficient electric heater and 4 GJ/MWh electricity equivalent), that is 9.7-13.9 MWh per day, or roughly 1.0-1.4 kWh per kg of LH2 produced. That number is material — for a 10 t/day LH2 plant it represents 4-6% of total liquefaction energy — and is the reason operators look hard at heat-integrated regeneration schemes (slipstream of cold-box reject gas for the purge, hot process gas from the synthesis gas preheater as the heating medium) before committing to a baseline electric-resistance regeneration design.

Co-adsorption of CO2 and hydrocarbons: the secondary loading that sneaks up on you

The single most common reason LH2 pre-purification 13X beds fail prematurely is not the water loading — operators learn to size for water on the first cycle — but the CO2 and light-hydrocarbon co-adsorption that accumulates across cycles when regeneration is not aggressive enough. CO2 binds to 13X with a heat of adsorption of about 42-48 kJ/mol (compared to 44-50 kJ/mol for water), which means CO2 is reversible at the same regeneration temperatures as water, but the equilibrium partial pressure of CO2 over 13X at 280-300 deg C is roughly 1-5 mbar — so the regeneration must reduce the CO2 partial pressure in the purge gas below 1 mbar to fully recover the bed. Light hydrocarbons (C1, C2, C3) have lower heats of adsorption (15-25 kJ/mol) and are more easily desorbed, but they accumulate during every adsorption cycle and require a small (5-10%) purge-gas flow at the end of the regeneration cycle to fully strip them. The cumulative effect across 200-400 cycles is a steady drift in bed capacity, with operators reporting a 1-2 wt% per month loss of effective water capacity that they mistakenly attribute to thermal degradation of the zeolite when the real cause is residual hydrocarbon and CO2 fouling. The fix is a periodic (every 30-60 days) deep regeneration at 320-340 deg C with a dry N2 purge flow for 12-18 hours, which restores the bed to within 0.5 wt% of the original loading.

Pre-13X bed configuration: amine scrubber, molecular gate, or hybrid

Almost every LH2 pre-purification train today has an upstream bulk-CO2 removal stage before the 13X bed, because trying to push all the CO2 onto the 13X cuts the water capacity roughly in half (CO2 occupies active sites that would otherwise hold water). The two dominant upstream technologies are:

• Amine scrubbing (aMDEA, MDEA-promoted): removes 95-99% of the CO2 from the feed gas at 30-70 deg C, reducing the inlet CO2 load to the 13X bed from typical 100-1000 ppmv down to 5-20 ppmv. Trade-off: amine carryover into the 13X bed is a known fouling mechanism that destroys 30-60% of water capacity within 20-40 cycles if the amine carryover exceeds 0.5 ppmv at the 13X inlet. The mitigation is a downstream KO drum and a wash-water section, but it adds CAPEX and a maintenance burden.
• Molecular gate (Carbon Molecular Sieve, CMS) layer above the 13X bed: physically separates CO2 from H2O at the kinetic-diameter level; CO2 (0.33 nm) is kinetically slow to enter the CMS pores and preferentially bypasses to a CO2-rich vent stream, while H2O (0.265 nm) enters freely. Trade-off: more complex bed design, requires careful pressure-equalization between CMS and 13X layers, and the CMS layer needs its own regeneration cycle at slightly different conditions than the 13X.

For green hydrogen (electrolysis-fed LH2 plants), the CO2 load is so low (typically <10 ppmv at the electrolyser outlet) that a single 13X bed with no upstream amine or CMS layer is often sufficient. For blue or grey hydrogen (SMR or ATR feed gas), the upstream amine scrubber is essentially mandatory. The bed sizing, regeneration cycle, and operating cost all change by 30-50% depending on which upstream configuration is chosen, so the 13X bed design is never standalone — it is always part of an integrated upstream-downstream package.

Start-up, commissioning, and the first-20-cycle capacity loss

A new 13X bed shipped from the manufacturer arrives with 1-2 wt% factory moisture content (from packaging in humid air) and a small amount of manufacturing dust on the bead surfaces. If the bed is loaded and put into service without a deep regeneration cycle, the first 3-5 adsorption cycles will show an apparent capacity that is 30-50% below design because the active sites are pre-loaded with manufacturing moisture. The standard commissioning procedure, drawn from EIGA IGC 121/14 and from the published commissioning manuals of major LH2 plant builders (Linde, Air Liquide, Iwatani), is:

1. Pre-load bed settling: load the 13X beads slowly through the top manway, vibrating the bed at 5-10 Hz for 10-15 minutes to settle the bed to design bulk density (typically 620-680 kg/m^3 for 4x8 mesh). Target a bed height within +/-2% of the design value.
2. Initial purge: pass dry N2 (<-40 deg C dew point) downward through the bed at 0.05-0.10 m/s superficial velocity for 4-6 hours to flush manufacturing dust and air from the void spaces.
3. First deep regeneration: heat the bed to 300-320 deg C at 1-2 deg C/min ramp, hold for 8-12 hours with a 200-400 Nm^3/h dry N2 purge, then cool to adsorption temperature at 2-3 deg C/min under continued N2 purge. This removes the factory moisture and conditions the bed for the first adsorption cycle.
4. First adsorption cycle: bring the feed gas in at 50% design flow for the first 4 hours, then ramp to 100% over the next 4 hours. Monitor outlet H2O with a chilled-mirror hygrometer and outlet CO2 with an IR analyzer; target <1 ppmv H2O and <2 ppmv CO2 throughout the cycle.
5. Baseline performance record: log pressure drop across the bed, outlet H2O, outlet CO2, and bed temperatures (top, middle, bottom) for the first 5 cycles. This baseline is the reference for the entire bed life and is used to detect capacity degradation, channeling, or upstream amine carryover.

Skipping steps 2 or 3 is the single most common cause of "my new 13X bed is underperforming" complaints in the first month of service. The factory moisture and dust are real, they are predictable, and they can be fully removed by following the commissioning procedure.

Common failure modes and how to detect them before they cost you a BOG surge

Beyond the standard capacity degradation that hits every adsorbent bed at end-of-life (typically 3-5 years for LH2 pre-purification service, sometimes 7-10 years in low-load, well-regenerated service), there are several failure modes specific to LH2 applications:

• Amine carryover from upstream scrubber: manifests as a rapid (within 20-40 cycles) 30-60% capacity loss, with a brown-yellow discoloration of the top 5-10 cm of the bed. Detection: weekly outlet H2O trend; if the breakthrough time drops 10% week-over-week and the bed top shows discoloration, suspect amine. Mitigation: inspect and overhaul the upstream amine scrubber KO drum.
• Hydrocarbon accumulation: manifests as a slow (1-2% per month) capacity loss, with no visible discoloration but a measurable (5-15%) increase in bed pressure drop. Detection: monthly pressure-drop measurement at design flow. Mitigation: deep regeneration at 320-340 deg C with dry N2 purge for 12-18 hours.
• Channeling at the vessel wall: manifests as early breakthrough at 10-30% of design cycle length, often after a thermal cycling event or a feed gas composition upset. Detection: multi-point temperature thermocouples in the bed (top, middle, bottom, plus one at the wall) showing early temperature rise at the wall thermocouple. Mitigation: full bed unload, vessel inspection, screen and distributor replacement, bed reload.
• Thermal shock cracking: manifests as fines generation and a step increase in post-bed particulate count. Detection: weekly outlet particulate count (target <1 mg/Nm^3). Mitigation: smooth the regeneration temperature ramp rate to 1-2 deg C/min and verify the heater duty is within design.

13X vs activated alumina vs 5A: which wins for LH2 pre-drying

This comparison comes up in nearly every LH2 plant design review. The trade-offs are:

13X is the dominant choice for LH2 pre-purification because the co-adsorption of CO2 is a feature (it removes both water and CO2 in one bed) and the higher working capacity at the actual operating temperature (-10 to -30 deg C) more than compensates for the slightly higher regeneration temperature. Activated alumina is sometimes used as a guard bed upstream of the 13X to remove the bulk water at lower regeneration cost, but the lower working capacity means more frequent regeneration cycles and a higher lifetime OPEX. 5A is reserved for CO2-free service (e.g., a hydrogen feed that has already been through a PSA CO2 removal stage) where the CO2 co-adsorption of 13X would be wasted.

Designing for the next 10 years: lifecycle cost and bed replacement planning

A 13X bed in LH2 pre-purification service has a design life of 3-5 years for high-cycle service (24-hour operation, 4-8 hour cycles) and 7-10 years for low-cycle service (12-hour operation, 12-24 hour cycles). The end-of-life signature is a steady decline in working capacity to 60-70% of design, accompanied by a 15-25% increase in bed pressure drop. At that point, full bed replacement is more cost-effective than continued regeneration, because the regeneration energy cost has increased to 1.5-2x the design value and the bed no longer meets the 1 ppmv H2O outlet target reliably.

For a 1500 kg 13X bed serving a 10,000 Nm^3/h LH2 plant, the lifecycle cost over a 5-year design life (assuming 3-4 full regeneration cycles per day, 330 operating days per year) breaks down approximately as:

• Initial fill cost: 1500 kg × USD 5/kg = USD 7,500.
• Annual regeneration energy cost: 50 GJ/day × 330 days × USD 25/GJ (electric heater equivalent) = USD 412,500/year.
• Annual purge gas cost (N2 at USD 0.05/Nm^3): 300 Nm^3/cycle × 3 cycles/day × 330 days × USD 0.05 = USD 14,850/year.
• Annual bed replacement reserve: USD 7,500 / 5 years = USD 1,500/year.
• Annual maintenance labour: 200 hours × USD 80/hour = USD 16,000/year.

Total annualized operating cost: roughly USD 445,000/year, of which 93% is regeneration energy. That single number — the regeneration energy cost — is why every LH2 plant operator is aggressively pursuing heat integration (using cold-box reject gas for the regeneration purge) and why new adsorbent developments in the low-temperature regeneration space (typically targeted at 200-250 deg C regeneration with maintained capacity) are watched so closely. For the operator, a 20% reduction in regeneration energy cost is worth more than any incremental improvement in equilibrium capacity.

Conclusion: getting the LH2 pre-purification design right the first time

The 13X molecular sieve pre-drying stage is the gatekeeper for every LH2 plant: it determines whether the cold box sees clean gas that lets the liquefier run at design efficiency, or whether it sees trace contaminants that nucleate as ice, plug vaporiser plates, and drive BOG rates up by 5-15%. The engineering decisions that matter most are (a) sizing the bed for the actual operating temperature (-10 to -30 deg C) rather than the brochure 25 deg C figure, (b) choosing 4x8 mesh over 8x12 mesh unless pressure drop is critically constrained, (c) committing to an L/D ratio of 3:1 to 4:1 rather than going shorter or taller, (d) budgeting regeneration energy at 1100-1250 kJ per kg of 13X per cycle, and (e) running the full commissioning procedure with at least one deep regeneration before the first adsorption cycle. Get all five right and the bed runs to its 5-year design life with a steady 10-12 hour cycle and a reliable <1 ppmv H2O outlet; cut any corner and the plant pays for it in BOG surcharges, premature bed replacement, or both.

Frequently Asked Questions

Next Steps

Need help sizing a 13X molecular sieve bed for a green or blue hydrogen LH2 pre-purification train, or specifying the upstream amine scrubber and ZnO guard bed to meet ISO 21010 and EIGA IGC 121/14 outlet targets? Talk to our application engineers directly — we ship ISO 9001-certified 13X in 4x8 and 8x12 mesh, with full CoA, attrition index, and equilibrium isotherm data on every drum.

• WhatsApp: +86 133 2522 2240
• Email: barry@aluminaworld.com
• Or request a free 1 kg sample + technical datasheet at /contact.html

Related Technical Articles

• Molecular Sieve 13X for Cryogenic Air Separation Pre-Purification
• Activated Alumina NH3 Cracking Residual Slip — Hydrogen Carrier
• Activated Alumina LNG Pre-Drying Bed Sizing
• Regeneration Cycle Getting Longer: 7 Root Causes Ranked
• Retrofitting an Existing Dryer to a New Adsorbent: 9 Checks

Featured Product

Molecular Sieve 13X (Sodium Form, X-Type Zeolite)

0.9 nm nominal pore opening, 22-25 wt% equilibrium water capacity at 25 deg C; co-adsorbs H2O and CO2 in a single bed for LH2 pre-purification, oxygen concentrators, and CO2 removal from natural gas.

View Product Details →