Molecular Sieve for Syngas Purification: CO2, CO, and H2S Removal in SMR and Partial Oxidation Hydrogen Plants

Why Syngas Purification Matters in Hydrogen Production

Hydrogen is the simplest molecule in the universe, but producing it at industrial scale is one of the most adsorbent-intensive processes in petrochemicals. The reason: hydrogen is made by reacting a hydrocarbon feedstock (natural gas, naphtha, refinery offgas, or coal) with steam and/or oxygen to produce a mixture called syngas, which is then purified to recover the H2. The syngas coming off a steam methane reformer (SMR) or a partial oxidation (POX) reactor contains roughly 70 to 80 percent H2, 15 to 25 percent CO2, 1 to 5 percent CO, 0.5 to 2 percent CH4, trace N2, and depending on the feedstock anywhere from less than 1 ppmv to several thousand ppmv of H2S plus COS.

To make 99.99 percent pure hydrogen for ammonia synthesis, fuel cells, or hydrotreating, the syngas must be cleaned up in three steps: shift conversion (CO + H2O -> CO2 + H2), acid gas removal, and final purification. The acid gas removal and final purification are almost universally done by pressure swing adsorption (PSA) on a bed of molecular sieve. This is the step where the right sieve specification saves millions of dollars in unplanned downtime and methanator catalyst replacement.

The choice of molecular sieve for syngas PSA is dominated by 13X, with supporting roles for 5A, 4A, and activated alumina. The 13X does the heavy lifting on CO2, H2S, and COS removal. 5A or 4A may be added as a polishing layer for cold-box-integrated plants. Activated alumina always sits at the front of the train as a water guard. The chemistry, capacity, and service life of each layer is what this guide covers in detail.

Syngas Composition by Source: Why It Matters for Sieve Selection

Before we go into sieve selection, you need to understand that syngas composition varies wildly by production route. A sieve specification that works perfectly for one feed may fail rapidly on another. The table below summarizes typical compositions from the four major industrial syngas sources:

The high CO2 partial pressure on POX and coal gasification feeds means a heavier sieve loading per Nm3 of product hydrogen. The high sulfur on refinery offgas means you need a layered bed with a ZnO or activated carbon guard upstream of the molecular sieve. For sweet SMR service from clean natural gas the design is straightforward: 13X primary, activated alumina pre-bed, done. The complexity scales with sulfur and contaminants in the feed.

The Chemistry: Why 13X is the Workhorse for Syngas PSA

13X is the sodium form of the faujasite (FAU) framework with a Si/Al ratio of about 1.2 and a pore opening of roughly 10 Angstrom. The pore is large enough to admit CO2 (3.3 Angstrom kinetic diameter), H2S (3.6 Angstrom), COS (kinetic diameter around 4.0 Angstrom), and even the larger mercaptans and thiophenes. The sodium cations at the SII and SIII sites inside the pore create a strong local electric field that preferentially adsorbs molecules with high quadrupole moments (CO2, N2) or polar bonds (H2S, COS, H2O).

On CO2 the static capacity of 13X at 25 degrees C and 250 mmHg partial pressure (a typical SMR PSA desorption condition) is 5.5 to 6.5 mmol/g, or roughly 24 to 28 weight percent. This is the highest equilibrium CO2 capacity of any commercial molecular sieve. By comparison, 5A delivers 3.5 to 4.5 mmol/g under the same conditions and activated carbon delivers 1.5 to 2.5 mmol/g. The capacity difference translates directly into bed size, cycle time, and hydrogen recovery.

The same sodium cation sites that bind CO2 also bind H2S, COS, and H2O. This is critical for syngas service because all three contaminants are usually present. The binding strength follows the order H2O >> COS > H2S > CO2 > CH4 > CO > H2, which means regeneration in a typical pressure-swing cycle desorbs H2 and CO first, then CH4, then CO2, then H2S, with COS and water staying adsorbed the longest. The practical consequence: H2S and COS slip past the bed only when the bed is heavily loaded and the cycle has to be shortened, which is the design alarm condition on every SMR PSA.

Why 13X wins over 5A and 4A in syngas service

5A has a 5 Angstrom pore opening that admits CO2 (3.3 Angstrom) and H2S (3.6 Angstrom) but excludes COS and most mercaptans. This means COS and organosulfur compounds slip past a 5A bed and poison the downstream methanator or ammonia synthesis catalyst. In SMR service where the only acid gas target is CO2, 5A can work in principle, but the bed is more sensitive to water and to contamination from minor feed upsets, and the cycle time has to be shortened to achieve the same purity. In practice, 5A is relegated to the polishing layer in modern SMR designs rather than the bulk CO2 removal bed.

4A has a 4 Angstrom pore opening. It admits water but excludes CO2 (3.3 Angstrom fits, but the diffusion is restricted) and excludes COS entirely. 4A is used almost exclusively as the final polishing layer where the design target is to drop water below 0.1 ppmv for cold-box integration. A 4A bed cannot handle bulk CO2 removal; the capacity is too low and the cycle time too short.

Activated carbon has a slit-shaped pore system in the 10 to 50 Angstrom range with no discrete cation sites. Adsorption is by van der Waals forces only, which makes carbon much weaker on CO2 than 13X. Carbon shines on heavier hydrocarbons and on H2S at low partial pressure, which is why it is the standard sulfur guard bed upstream of an amine or ZnO system. But as the primary CO2 adsorber in syngas PSA, carbon needs 2 to 3 times more bed volume than 13X. In high-pressure SMR service this is not economic.

Why not LiLSX for syngas PSA?

LiLSX is the lithium-exchanged form of the X-type framework used in medical oxygen concentrators because of its exceptionally high N2 capacity. LiLSX has been proposed for syngas PSA by a few academic groups but has not been adopted industrially for two reasons. First, the LiLSX capacity advantage over 13X for CO2 is only 10 to 15 percent, not the 2 to 3 times advantage seen for N2 over O2. Second, LiLSX is much more expensive than 13X (5 to 8 times the price per kg), and the small capacity gain does not justify the CAPEX premium. Third, LiLSX is far more sensitive to water than 13X, requiring more elaborate pre-bed design and tighter regeneration control. For the foreseeable future 13X remains the standard for syngas PSA, with LiLSX reserved for oxygen separation and a few niche gas purification applications.

Side-by-Side Adsorption Data: 13X vs 5A vs 4A vs Activated Carbon

The numbers below are at 25 degrees C and standard test conditions. They are typical values from Aluminaworld in-house testing cross-checked against published UOP, CECA, and Air Products reference curves. All values are static equilibrium capacities measured on a Rubotherm magnetic suspension balance or a Micromeritics ASAP 2020 volumetric unit.

The takeaway: 13X does everything. 5A and 4A are used as polishing layers where residual CO2 must drop below 10 ppmv for cold-box integration or fuel-cell-grade H2. Activated carbon is reserved for low-pressure sulfur guard duty in front of the 13X, never as the primary CO2 adsorber in high-pressure syngas service.

Layered Bed Design: How the SMR PSA Train Is Built

A modern SMR hydrogen plant PSA train is not a single bed of one sieve. It is a layered stack, with each layer selected for a specific contaminant. The most common configuration is:

1. Bottom layer (water guard): 15 to 20 percent of bed height, loaded with 3 to 5 mm activated alumina beads. Removes entrained water from upstream knockout drums and from the shift converter effluent. Typical loading 4 to 6 tonnes per bed in a 100,000 Nm3 per day plant.
2. Middle layer (bulk CO2 removal): 50 to 60 percent of bed height, loaded with 1.6 to 2.5 mm 13X-HG beads. This is the workhorse layer. Removes 80 to 90 percent of the CO2 and most of the H2S, COS, and mercaptans. Typical loading 14 to 20 tonnes per bed.
3. Top layer (polishing): 20 to 30 percent of bed height, loaded with either 5A or 4A beads in the 1.6 to 2.5 mm range. Drops residual CO2 below 50 to 100 ppmv and water below 1 ppmv. Critical for plants with downstream methanator or cold box. Typical loading 4 to 6 tonnes per bed.

The bed heights are calculated from the working capacity of each layer at the operating pressure, with a 1.2 to 1.5 safety factor on the mass transfer zone (MTZ) length. The MTZ is the section of bed where the contaminant front is still moving; beds must be long enough that the MTZ does not reach the bed exit during the adsorption half-cycle. For a 13X bulk CO2 layer in a sweet SMR the MTZ is 0.3 to 0.5 m, so total bed height of 4 to 7 m provides 2 to 4 m of safety margin.

Bed sizing example for a 100,000 Nm3/day SMR plant

Let us run the numbers. A 100,000 Nm3 per day SMR plant runs at 28 bar adsorption and 1.5 bar desorption with a 10-minute cycle. Syngas feed rate is approximately 130,000 Nm3 per day (product recovery is 77 percent). Each adsorption bed processes 4,500 Nm3 per cycle. Required CO2 working capacity per cycle is approximately 1,200 Nm3 (about 26 percent of feed).

13X-HG working capacity at 28 bar adsorption / 1.5 bar desorption is approximately 9.0 to 10.0 mmol/g, or 0.40 to 0.44 Nm3 CO2 per kg of sieve. To adsorb 1,200 Nm3 CO2 you need 2,700 to 3,000 kg of 13X per bed. With 6 parallel beds in the train (4-bed, 6-bed, 8-bed, or 12-bed configurations are common), the 13X inventory is 16 to 18 tonnes for the bulk CO2 layer alone. Add the activated alumina water guard (4 tonnes) and the 5A polishing layer (5 tonnes) and total sieve per bed is 25 to 27 tonnes. Across 6 beds: 150 to 160 tonnes of sieve per train. At current spot prices of 4.5 to 6.5 USD per kg for 13X-HG in bulk, the sieve alone is 700,000 to 1,000,000 USD per train.

Mass transfer zone length calculation

The mass transfer zone (MTZ) is the section of bed where the contaminant concentration is between 5 percent and 95 percent of the inlet value. For a properly designed bed, the MTZ must not reach the bed exit during the adsorption half-cycle, otherwise contaminants break through. The MTZ length is calculated from:

L_MTZ = (u x t_cycle) / (RHO_b x W x 1000)

Where u is the superficial gas velocity in m/s, t_cycle is the adsorption time in seconds, RHO_b is the bulk density of the sieve in kg/m3, and W is the working capacity in kg adsorbate per kg sieve. For the SMR example above: u = 0.25 m/s, t_cycle = 300 s, RHO_b = 640 kg/m3, W = 0.09 kg/kg. So L_MTZ = (0.25 x 300) / (640 x 0.09 x 1000) = 75 / 57600 = 0.0013 m = 1.3 mm. This is a very short MTZ, which is good news: it means the 13X bed is highly efficient and the bed length is set by total capacity, not by MTZ.

For larger PSA units operating at lower velocity (0.10 to 0.15 m/s) and longer cycle times (600 to 900 s), the MTZ can grow to 5 to 15 mm. Total bed length is set as MTZ plus a safety factor of 1.5 to 2.0 times the MTZ, plus the equilibrium loading section. Typical bed length to diameter (L/D) ratios for syngas PSA are 1.5 to 2.5. Lower L/D means more end-effects and shorter bed life; higher L/D means higher vessel cost without capacity benefit.

Pressure drop through the bed

Bed pressure drop is calculated using the Ergun equation, which balances viscous and inertial losses against the bed geometry. For the 13X bulk layer at SMR conditions:

dP/L = 150 x (1 - eps)^2 / eps^3 x mu x u / dp^2 + 1.75 x (1 - eps) / eps^3 x rho x u^2 / dp

Where mu is gas viscosity (about 1.8 x 10^-5 Pa s for syngas at 25 C), u is superficial velocity, eps is bed void fraction (about 0.40 to 0.42 for 1.6 to 2.5 mm beads), dp is bead diameter, and rho is gas density. For our SMR example: dP/L = 150 x 0.348 / 0.068 x 1.8e-5 x 0.25 / (2.0e-3)^2 + 1.75 x 0.6 / 0.068 x 18 x 0.0625 / 0.002 = 1600 + 7300 = 8900 Pa/m = 8.9 kPa/m of bed. Over a 4 m bed, total pressure drop is 36 kPa or about 0.36 bar. This is acceptable: typical design targets are 50 to 100 kPa total bed pressure drop at full operating conditions.

If pressure drop exceeds 100 kPa in service, the most likely cause is dust accumulation from sieve attrition. The fix is to inspect the bed top for dust, vacuum out the top 50 to 100 mm of fines, and replace with fresh sieve. Do not top up with new sieve on top of an existing bed without removing the fines first; the new sieve will simply mix with the dust and the problem recurs within months.

H2S Management: The Leading Cause of Premature Sieve Failure

The number one reason molecular sieve beds in syngas PSA service fail early is H2S poisoning. H2S binds to the 13X cation sites with a heat of adsorption of roughly 35 to 40 kJ/mol, compared to 20 to 25 kJ/mol for CO2. This means H2S is preferentially adsorbed over CO2, occupies cation sites that would otherwise hold CO2, and is harder to desorb. Over thousands of cycles, residual H2S accumulates on the bed and the working CO2 capacity drops.

On a clean sweet SMR with less than 1 ppmv H2S in the feed, the residual H2S build-up is negligible and 13X beds last 10 to 15 years. On a refinery offgas feed with 100 to 500 ppmv H2S, even with a ZnO guard bed ahead of the PSA, the residual H2S in the 1 to 5 ppmv range will cut sieve life to 5 to 8 years. On a coal gasification syngas with 500 to 5000 ppmv H2S upstream of an amine scrubber, expect 3 to 5 years. There is no field fix once H2S has accumulated on the sieve. The bed must be replaced.

Guard bed options for H2S

The standard guard bed upstream of a syngas PSA is a ZnO fixed bed. ZnO reacts with H2S to form ZnS at temperatures of 200 to 400 degrees C, dropping the H2S from 100 to 1000 ppmv down to less than 1 ppmv. ZnO beds have a sulfur capacity of 15 to 25 weight percent before breakthrough. The bed is replaced, not regenerated. A typical 100,000 Nm3 per day SMR with 50 ppmv H2S in the feed uses 6 to 10 tonnes of ZnO per year.

An alternative is impregnated activated carbon. Carbon impregnated with copper or zinc salts can drop H2S from 100 ppmv to less than 0.1 ppmv at room temperature. The carbon bed is regenerated with hot steam (200 to 300 degrees C) every 24 to 72 hours, which makes it suitable for cyclic service but adds steam infrastructure. Carbon guard beds are more common in refinery offgas plants where the H2S is intermittent and the cost of ZnO replacement is high.

A third option is a water-wash plus amine absorber upstream of the PSA. The amine (usually MDEA, monoethanolamine, or a formulated blend) drops H2S to less than 4 ppmv in a single contactor stage and to less than 0.5 ppmv in two stages. This is the most expensive option CAPEX-wise but is often justified for very sour syngas streams (more than 1 percent H2S) where ZnO consumption would be uneconomic.

Diagnosing H2S breakthrough in service

Operators can detect the early stages of H2S breakthrough by watching three key PSA operating parameters. First, the cycle time has to be shortened to maintain product purity. If the cycle time drops 10 to 15 percent over 12 to 24 months with no change in feed conditions, H2S accumulation is the most likely cause. Second, the bed temperature profile during regeneration shifts: the desorption temperature front moves down the bed more slowly because H2S desorbs later than CO2. Third, the H2S slip on the product side rises above 0.1 ppmv if the methanator or downstream catalyst has a continuous H2S analyzer.

If any of these symptoms appear, take a representative sieve sample from the top of the bed (where H2S accumulates first) and send it for laboratory analysis. A simple H2S loading measurement by thermogravimetric analysis (TGA) with mass spec detection will tell you how much sulfur is on the bed. Loading above 0.5 weight percent sulfur indicates the bed is approaching end-of-life. Loading above 1.0 weight percent means the bed needs to be replaced at the next planned turnaround.

Recovering H2S-poisoned sieve: is it possible?

In theory, an H2S-loaded 13X bed can be regenerated with high-temperature steam (250 to 350 degrees C) plus a reducing gas mixture (5 percent H2 in N2) to convert adsorbed H2S to elemental sulfur and drive it off the bed. In practice this is rarely done at industrial scale because the high temperature degrades the zeolite binder, the reducing gas creates explosive atmospheres, and the regenerated capacity is rarely more than 60 to 70 percent of fresh. The economic case for off-site regeneration is weak. Operators replace, not regenerate, poisoned beds.

The only realistic mitigation strategy for an H2S-poisoned bed is preventive: install a ZnO or activated carbon guard upstream of the PSA, monitor the H2S slip continuously, and replace the guard bed material at the first sign of breakthrough. The cost of prevention is 50,000 to 150,000 USD per year. The cost of an unplanned sieve replacement is 500,000 to 4,000,000 USD per event. The math is obvious.

Cold Box Integration and the Final Polishing Bed

Hydrogen plants that include a downstream cryogenic cold box (typical of LNG production, hydrogen liquefaction, or helium recovery from natural gas) have an additional sieve specification requirement: the final polishing bed must drop CO plus CO2 to less than 1 ppmv and water to less than 0.1 ppmv. Otherwise trace CO2 and water freeze in the cold box heat exchangers and cause unplanned shutdowns.

Cold-box pluggage is the single most expensive unplanned shutdown in a hydrogen plant. Restarting a plugged cold box requires a 24 to 72 hour warm-up cycle, during which the entire plant is offline. Production loss at a 100,000 Nm3 per day plant is approximately 100,000 to 300,000 USD per day depending on regional hydrogen prices. The fix is a final polishing bed of 4A or 5A sieve after the main 13X layer.

For plants with a downstream methanator the final polishing layer must also drop CO2 to less than 10 ppmv, because CO2 poisons the nickel methanation catalyst. The 4A or 5A polishing layer handles this. If you operate a fuel-cell-grade H2 plant (ISO 14687 purity standard), the final polishing layer is mandatory and you should plan on a 13X + 5A + 4A three-layer bed.

Regeneration Cycle Design: Pressure Swing vs Temperature Swing

Syngas PSA regeneration is done by pressure swing, not by temperature swing. The bed is depressurized from the adsorption pressure (typically 25 to 35 bar) to the desorption pressure (typically 1 to 3 bar) and then purged with a slipstream of product hydrogen at the desorption pressure. The desorbed CO2, H2S, and water are swept out of the bed to the tail gas system, where they are recycled as reformer fuel.

The standard cycle has six steps: adsorption, pressure equalization (top), pressure equalization (bottom), depressurization, purge, and repressurization. Modern SMR designs include two or three pressure equalization steps to maximize hydrogen recovery, which reaches 85 to 92 percent in a well-tuned 6-bed or 8-bed system. Recovery below 80 percent indicates sieve capacity degradation or cycle time misconfiguration.

The purge step uses product hydrogen at 1.5 bar, typically 5 to 15 percent of the product flow rate. This is the main hydrogen loss mechanism in PSA. Modern SMR designs minimize this loss through the equalization steps, but you always have some hydrogen going to the tail gas system. The tail gas is recycled as reformer fuel, so the energy is not wasted, but it does reduce net hydrogen yield.

Hydrogen recovery formulas

The hydrogen recovery of a syngas PSA is defined as the ratio of pure H2 in the product stream to total H2 in the syngas feed. For a sweet SMR with 6-bed configuration, recovery is calculated as:

R_H2 = (V_product x x_H2_product) / (V_feed x x_H2_feed) x 100%

Where V_product and V_feed are the volume flow rates (Nm3/h) and x_H2 are the H2 mole fractions. Typical values are 78 to 92 percent. The loss is primarily to the tail gas during purge and equalization steps.

For a 6-bed system with two equalization steps, the recovery is approximately:

R_H2 approx = 1 - (n_purge + n_equalization) / n_adsorption

Where n_purge is the moles of H2 used in the purge step, n_equalization is the moles lost during equalization transfers, and n_adsorption is the moles of H2 in the feed during the adsorption step. The purge gas requirement is roughly 0.05 to 0.15 Nm3 per Nm3 of product H2, depending on the desorption pressure and the purge-to-feed ratio.

A practical formula used by SMR designers for first-pass recovery estimation is:

R_H2 approx = (P_ads - P_des) / (P_ads - 1) x eta

Where P_ads is the adsorption pressure in bar, P_des is the desorption pressure in bar, and eta is the cycle efficiency factor (0.85 to 0.95 for well-tuned systems). For a typical 28 bar adsorption / 1.5 bar desorption system, this gives R_H2 approx = (28-1.5)/(28-1) x 0.90 = 27/27 x 0.90 = 0.90 or 90 percent. Field measurements typically come in 2 to 5 percent below the theoretical formula due to real-world inefficiencies in valve switching and pressure equalization timing.

How sieve degradation affects recovery

As the sieve ages and CO2 working capacity drops, the cycle time has to be shortened to maintain product purity. Shorter cycle time means more frequent regeneration, which means more purge gas used per unit of product H2. Net result: recovery drops 5 to 15 percent over the sieve life. A 90 percent recovery fresh sieve may drop to 78 percent recovery after 10 years in service. This is the single biggest operating cost driver for sieve replacement and the reason operators track recovery as a leading indicator of sieve health.

Aluminaworld 13X-HG Specifications for Syngas Service

For engineers ready to specify a sieve, here is the data sheet our hydrogen customers use:

Full lot-level Certificate of Analysis is provided with every shipment, including CO2 capacity, H2O capacity, particle size distribution, attrition, crush strength, and LOI (loss on ignition). We also offer third-party SGS inspection on request for procurement contracts.

Matched Activated Alumina Pre-Bed

Every 13X bed in syngas service must be paired with an upstream activated alumina pre-bed. The pre-bed removes the bulk of the water from the syngas, which protects the 13X from liquid water damage and reduces the regenerator duty on the 13X layer. Activated alumina for syngas pre-bed service is typically the 3 to 5 mm or 5 to 8 mm bead grade, with surface area of 320 to 380 m2/g and static H2O capacity of 17 to 22 weight percent.

Aluminaworld supplies matched activated alumina in the AW-300 series, with three bead size options (2 to 5 mm, 3 to 5 mm, 5 to 8 mm) and two purity grades (standard 93 percent Al2O3 and high-purity 95 percent Al2O3). For syngas pre-bed service the standard 93 percent grade is sufficient. The pre-bed is sized at 15 to 25 percent of the 13X layer height, depending on feed water content.

On a sweet SMR with 200 to 500 ppmv water in the syngas (after the shift converter and cooler), the pre-bed removes 95 to 99 percent of the water and the 13X sees less than 5 ppmv. On a wet refinery offgas stream with 1000 to 5000 ppmv water, the pre-bed must be larger and may need periodic in-situ regeneration with hot nitrogen (180 to 250 degrees C).

7 Common Failure Modes in Syngas PSA Molecular Sieve Beds

From field service records across more than 40 syngas PSA installations in 22 countries, the most common failure modes are well documented. If you operate a hydrogen plant PSA, watch for these seven patterns:

1. H2S accumulation on 13X. The leading cause of premature bed replacement. ZnO guard bed exhaustion or bypass valve failure allows H2S to reach the 13X. Capacity drops over 6 to 24 months. Symptom: cycle time has to be shortened to maintain product purity. Fix: replace 13X and inspect the ZnO system.
2. Liquid water carryover. Knockout drum failure or sudden process upset sends liquid water into the sieve bed. The water dissolves the zeolite binder and causes channeling and high pressure drop. Symptom: bed pressure drop doubles in days. Fix: replace bed, fix knockout drum, install better level control.
3. Hydrocarbon fouling from refinery offgas. Heavy hydrocarbons (C6+) condense in the bed and polymerize under the heat of adsorption. The bed turns yellow-brown and capacity drops. Symptom: gradual capacity loss over 12 to 24 months. Fix: install activated carbon guard bed upstream, replace sieve.
4. COS breakthrough. Carbonyl sulfide slips past the 13X if the bed is overloaded and the cycle time is too long. COS poisons downstream methanator or ammonia synthesis catalyst. Symptom: methanator nickel catalyst deactivates rapidly. Fix: shorten cycle time, add 5A polishing bed, replace sieve if COS slip persists.
5. Bed settlement and channeling. Mechanical vibration or thermal cycling causes the bed to settle, leaving voids at the top. Syngas bypasses the bed through the void and contaminates the product. Symptom: product purity drifts downward over months. Fix: top up the bed with fresh sieve, install bed-lift piping.
6. Sieve dusting from poor attrition resistance. Low-quality sieve with attrition above 0.5 weight percent generates dust that carries downstream and fouls piping, valves, and the methanator. Symptom: valve seat leakage, pressure drop oscillations. Fix: replace with high-quality low-dust sieve (attrition below 0.05 wt%).
7. Sulfur absorption bed cross-contamination. During turnaround or commissioning, sulfur-containing gases are mistakenly introduced into the 13X bed. Capacity drops irreversibly. Symptom: capacity loss immediately after startup. Fix: replace bed, improve commissioning procedures.

Most of these failures are preventable with proper specification, proper commissioning, and regular operating discipline. The annual cost of prevention (proper pre-bed design, regular sieve sampling, scheduled valve maintenance) is 50,000 to 200,000 USD per year. The cost of unplanned sieve replacement is 500,000 to 4,000,000 USD per event. The math is obvious.

5 Industrial Case Studies: Syngas PSA in the Field

The following case studies are summarized from real installations in Aluminaworld's customer base. Customer names are omitted but the technical details are accurate. Where data is anonymized for confidentiality, the engineering logic and the field outcomes are real.

Case Study 1: Sweet SMR in Saudi Arabia (Aramco-class), 200,000 Nm3/day

A 200,000 Nm3 per day SMR plant in eastern Saudi Arabia running on clean natural gas feed (less than 0.5 ppmv H2S in the feed gas) commissioned in 2010 with 13X-HG as the primary PSA bed. The original 13X load of 320 tonnes has been in service for 16 years as of this writing. Annual sieve sampling shows 91 percent of initial CO2 capacity remaining in 2026. Expected end-of-life is 2030 at the 80 percent capacity threshold. Total sieve replacement cost over the 20-year design life will be approximately 1.4 million USD for the initial load plus a mid-life top-up of 80 tonnes in 2030. By contrast, a competitor's 13X load in the same service was replaced after 9 years because of higher initial fines generation.

Case Study 2: Refinery offgas H2 plant in Texas, 75,000 Nm3/day

A 75,000 Nm3 per day hydrogen plant running on catalytic reformer offgas (350 ppmv H2S upstream, 2 ppmv downstream of an amine contactor) commissioned in 2014 with a layered bed of activated alumina + 13X-HG + 5A. The plant had two unscheduled shutdowns in 2017 and 2019 due to H2S breakthrough when the amine contactor level control failed. Both events required partial sieve replacement (40 tonnes each time). After the second event, the operating team installed a redundant ZnO guard bed ahead of the amine contactor. No further sieve failures in 7 years of subsequent operation. Service life on the current 13X load is on track for 10 years.

Case Study 3: Coal gasification H2 plant in China (Sinopec-class), 300,000 Nm3/day

A 300,000 Nm3 per day coal-to-hydrogen plant in northern China commissioned in 2018 with syngas at 50 bar and 1500 ppmv H2S upstream of a Rectisol wash. Rectisol drops the H2S to less than 0.1 ppmv, which is the cleanest syngas stream in the global hydrogen industry. The 13X-HG PSA beds (480 tonnes total inventory) have been in service for 8 years with no measurable capacity loss. Annual sieve sampling shows 97 percent of initial CO2 capacity. This represents the gold standard for sieve life in syngas service and demonstrates that with proper upstream sulfur removal, molecular sieve life is essentially limited by mechanical attrition rather than chemical degradation.

Case Study 4: POX H2 plant in Germany, 50,000 Nm3/day, cold box integrated

A 50,000 Nm3 per day partial oxidation hydrogen plant in Germany with a downstream cryogenic cold box for helium recovery. Commissioned in 2012 with a layered bed of activated alumina + 13X-HG + 5A + 4A. The 4A top layer was added specifically to drop CO + CO2 below 1 ppmv to prevent cold-box pluggage. The plant has had zero cold-box pluggage incidents in 14 years of operation. A neighboring plant without the 4A polishing layer had three cold-box pluggage events in the same period, each costing 250,000 to 400,000 EUR in lost production. The 4A layer added 60,000 EUR to the original sieve cost. The payback was inside the first year of operation.

Case Study 5: Biogas-to-H2 pilot in Brazil, 2,000 Nm3/day

A 2000 Nm3 per day biogas-to-hydrogen pilot in southern Brazil commissioned in 2023. Biogas from sugarcane vinasse digestion, 60 percent CH4, 35 percent CO2, trace H2S (less than 50 ppmv). The pilot uses a steam reformer followed by a small PSA train with 8 tonnes of 13X. The H2S guard is impregnated activated carbon. After 30 months of operation the 13X has lost 12 percent of initial CO2 capacity, which is faster than expected. Root cause analysis identified intermittent H2S spikes during feedstock variability. The fix in progress: install a second-stage activated carbon guard and a continuous H2S analyzer with auto-shutdown. Lesson learned: even trace H2S in biogas feed requires robust guard bed design.

Case Study 6: Ammonia plant H2 recovery in Indonesia, 120,000 Nm3/day

A 120,000 Nm3 per day ammonia synthesis plant in Indonesia uses a partial oxidation reformer followed by a 13X PSA train to recover H2 from the secondary shift converter offgas. The syngas is very wet (3000 ppmv water) and contains 200 ppmv H2S plus 30 ppmv COS after the Rectisol unit. The PSA was commissioned in 2015 with activated alumina + 13X-HG + 5A layered bed design. After 11 years of operation the 13X-HG beds have lost 18 percent of initial CO2 capacity. The team has scheduled a partial top-up of 35 tonnes in the 2027 turnaround. Total sieve spend over the 11-year period is approximately 1.6 million USD including the partial top-up. By contrast, a neighboring plant using a competitor's 13X had to do a full replacement (95 tonnes) in 2020 after only 6 years, due to sulfur accumulation and dust generation. The cost difference: 800,000 USD in favor of the high-quality 13X-HG over the same 11-year period.

Case Study 7: Hydrogen liquefaction plant in South Korea, 30,000 Nm3/day, cold box integration

A 30,000 Nm3 per day liquid hydrogen plant in South Korea uses a SMR followed by a 13X PSA train and a cryogenic cold box to liquefy the product H2 for transport. The cold box operates at -253 degrees C and cannot tolerate any CO2 or water slip. The PSA train uses 60 tonnes of layered bed: activated alumina + 13X-HG + 5A + 4A. The 4A polishing layer is critical - it drops the CO + CO2 below 1 ppmv and water below 0.1 ppmv required by the cold box. The plant has been in service for 7 years without a cold-box pluggage incident. Annual sieve sampling shows 96 percent of initial capacity remaining. The lesson: cold-box-integrated hydrogen plants must include the 4A polishing layer, period. The 80,000 USD incremental sieve cost is trivial compared to the 1 to 2 million USD cost of a single cold-box pluggage event.

Commissioning and Startup: 10 Steps to a Healthy First-Year PSA

A new molecular sieve bed installation is only as good as the commissioning process. Skipping steps or rushing the activation leads to under-utilized capacity, channeling, and a 20 to 30 percent first-year capacity loss that never recovers. The 10 steps below are derived from Aluminaworld commissioning experience on 40+ syngas PSA installations in 22 countries.

1. Pre-loading inspection. Inspect every PSA vessel internal for welding debris, loose bolts, and damaged distributor screens before loading sieve. A single loose bolt can damage thousands of dollars of sieve beads and cause channeling.
2. Sieve receiving inspection. Collect CoA from every drum. Sample 1 drum per 10 for verification testing: PSD, CO2 capacity, attrition. Verify against the procurement specification. Reject any drum out of spec.
3. Loading weather conditions. Load sieve only when the relative humidity is below 60 percent and the ambient temperature is between 5 and 35 degrees C. Loading in high humidity picks up 1 to 3 weight percent water before the sieve ever sees feed gas, which permanently reduces capacity.
4. Drop height limit. Use a loading chute with a maximum drop height of 1 meter from the sieve to the bed surface. Higher drop heights generate fines that will migrate to the bottom of the bed and cause high pressure drop.
5. Layered loading sequence. Load the layers in the correct order: bottom (water guard activated alumina) first, then bulk CO2 layer (13X-HG), then top polishing layer (5A or 4A). Use a layer separator mesh between each layer to prevent mixing.
6. Bed leveling. After each layer, level the surface with a manual screed or a vibrating plate. Avoid cone-shaped or pyramid-shaped bed surfaces, which cause channeling and premature breakthrough.
7. Initial activation. Heat the bed to 200 to 250 degrees C with hot nitrogen (or hydrogen) for 12 to 24 hours before introducing feed syngas. This removes any residual moisture picked up during loading. Skipping activation leaves 2 to 4 weight percent water in the bed, which reduces CO2 capacity by 15 to 20 percent.
8. First adsorption cycle. Start the PSA cycle at 50 percent of design adsorption pressure and gradually ramp to full pressure over 4 to 8 hours. Sudden pressurization shock-loads the bed and generates fines.
9. Performance verification. After 72 hours of steady-state operation, measure CO2 slip on the product side and H2 recovery. Compare to design specifications. Investigate any deviation greater than 10 percent before declaring the bed in service.
10. First-year sampling program. Sample the bed at 6 months and 12 months. Pull representative samples from the top, middle, and bottom of each layer. Test for CO2 capacity, H2O capacity, sulfur loading, and fines content. Compare against the fresh sieve baseline. This gives you a degradation curve to project end-of-life.

Plants that follow these 10 steps consistently achieve 90+ percent of design capacity in year one and project sieve life of 12+ years. Plants that skip the activation step or load in high humidity are typically at 70 percent of design capacity by year two and need sieve replacement at year 7 instead of year 15.

Standards and Specifications Governing Syngas PSA Sieves

Syngas PSA molecular sieve selection and operation is governed by a stack of international standards. The most relevant for procurement and operations engineers are:

• ISO 9001:2015 - Quality management system requirements for sieve manufacturers. All Aluminaworld sieve products are produced under ISO 9001 with full lot traceability.
• ASTM D5028 - Standard test method for determination of particle size of activated carbon and molecular sieve by sieve analysis. Used for sieve PSD verification.
• ASTM D4058 - Standard test method for attrition of activated carbon and molecular sieve. Used for sieve attrition loss verification.
• ASTM D4179 - Standard test method for single pellet crush strength of formed catalyst and molecular sieve. Used for crush strength verification.
• ISO 9277 - Determination of the specific surface area of solids by gas adsorption - BET method. Used for surface area verification of activated alumina and sieve support.
• ISO 13320 - Particle size analysis - Laser diffraction methods. Used for sub-millimeter sieve PSD.
• ISO 14687 - Hydrogen fuel quality specification for fuel cell vehicles. Defines the purity requirements that drive the polishing bed specification.
• ISO 8573-1 - Compressed air purity specification. Reference for inlet air quality if the hydrogen plant uses an air-driven compressor in the PSA.
• ASME BPVC Section VIII - Boiler and Pressure Vessel Code, rules for pressure vessel design. Governs the design of the PSA vessels themselves.
• PED 2014/68/EU - European Pressure Equipment Directive. Mandatory for any European plant.
• GB/T 6286 - Chinese national standard for molecular sieve determination of static water adsorption. Reference for Chinese-spec procurement.
• GB/T 34540 - Chinese national standard for molecular sieve attrition test. Used for domestic Chinese procurement.
• IEC 61511 - Functional safety standard for safety instrumented systems. Applies to PSA cycle control and emergency shutdown.
• API 571 - API damage mechanisms for fixed equipment in refineries. Reference for failure mode analysis of PSA vessels and internals.

When you write a sieve procurement specification, reference these standards by name. Most sieve suppliers (including Aluminaworld) can quote against a specification that cites ASTM and ISO test methods for every property listed on the data sheet.

Vendor Selection Checklist: 12 Questions to Ask Before You Buy

When you are evaluating molecular sieve suppliers for a syngas PSA installation, use this 12-question checklist. The answers tell you whether the supplier understands your application or is just selling commodity product.

1. What is the static CO2 capacity at 250 mmHg and 25 C? (Acceptable: 5.5 mmol/g minimum for 13X-HG.)
2. What is the attrition loss by ASTM D4058? (Acceptable: below 0.05 wt% for PSA-grade material.)
3. What is the single-bead crush strength by ASTM D4179? (Acceptable: above 35 N/bead for 1.6 to 2.5 mm beads.)
4. Can you supply lot-level CoA with PSD, capacity, attrition, and crush data per ASTM test methods?
5. What is the particle size uniformity? (Acceptable: above 95% within specified range.)
6. Can you match the bead size to the existing bed geometry (1.6 to 2.5 mm or 2.5 to 5.0 mm)?
7. What is the recommended pre-bed grade of activated alumina, and how do I size the pre-bed layer?
8. What is the recommended layered bed design for sour syngas (more than 100 ppmv H2S)?
9. Can you provide case studies from similar installations (SMR or POX with comparable capacity)?
10. What is the MOQ and lead time for production orders, and what is the FOB port?
11. Do you offer on-site commissioning support and sieve performance sampling?
12. What is the shelf life in unopened drums, and what is the storage condition recommendation?

A supplier who can answer all twelve confidently has done this work before. A supplier who can answer fewer than eight should be supplemented with a more experienced vendor or a technical consultant. Sieve specification is too important to leave to a procurement agent.

Cost Economics: Why Cheap Sieve is the Most Expensive Mistake

Let us run the TCO numbers for a 100,000 Nm3 per day SMR hydrogen plant over a 20-year design life. Two scenarios: (a) buy the cheapest available 13X at 3.50 USD/kg, and (b) buy a high-quality 13X-HG at 5.50 USD/kg. Both are technically 13X; the difference is in purity, uniformity, and dust generation.

The conclusion: spending 300,000 USD more on the premium sieve saves 3.6 million USD over the 20-year life of the plant. The premium sieve pays back inside the first 18 months from the hydrogen recovery improvement alone. This is why procurement agents who buy on lowest-bidder price lose money for their plants in the long run.

Future of Syngas PSA: Carbon Capture Implications

The biggest near-term change in syngas purification is the rise of carbon capture and sequestration (CCS) requirements. New hydrogen plants in the EU, UK, Canada, and several US states are required to capture 90 to 95 percent of the CO2 in the syngas stream. This drives two changes to the PSA design:

First, the PSA tail gas (the CO2-rich desorption stream) is no longer routed back to the reformer as fuel. Instead it is compressed and sent to a sequestration well or used for enhanced oil recovery. This eliminates the reformer fuel credit but adds compression cost. The PSA cycle does not change.

Second, the PSA cycle is being optimized for higher CO2 capture rate rather than higher H2 recovery. Modern CCS-designed SMRs run at 78 to 82 percent H2 recovery instead of the 86 to 90 percent typical of conventional SMRs. This means slightly larger sieve beds and slightly more sieve inventory per Nm3 of product H2. The sieve specification itself does not change much, but the quantity goes up by 5 to 10 percent.

For green hydrogen plants using electrolyzers rather than reformers, the CO2 capture question is moot - there is no CO2 in the feed. But for blue hydrogen (SMR with CCS) and pink hydrogen (electrolysis with nuclear), the CO2 question is increasingly important. Expect sieve suppliers to develop new grades optimized for CCS-specific cycle profiles over the next 5 years. Aluminaworld has a CCS-grade 13X in development for delivery in 2027.

Conclusion: Match the Sieve to the Syngas, Not the Other Way Around

Syngas purification in SMR and POX hydrogen plants is dominated by 13X molecular sieve, with supporting roles for activated alumina, 5A, and 4A. The sieve specification drives both the upfront CAPEX and the 20-year operating cost of the plant. Get the specification right and the sieve train runs for 15+ years with a single mid-life top-up. Get it wrong and you replace the bed every 5 to 8 years while poisoning the downstream methanator.

The three highest-impact decisions in syngas sieve selection are: (1) match the bead size to the bed geometry, (2) add the right guard bed upstream of the 13X for the sulfur and water content of your feed, and (3) decide whether you need a polishing layer for cold-box integration or fuel-cell-grade product. If you get those three right, the rest is execution.

For engineers who want to go deeper, the Aluminaworld technical team has direct experience with syngas PSA installations in 22 countries across SMR, POX, coal gasification, and refinery offgas feedstocks. We can help you size the layered bed, recommend the right guard system, and quote the matched sieve train with full CoA and on-site commissioning support.

Next Steps for Your Syngas PSA Project

If you are designing a new hydrogen plant or retrofitting an existing PSA train, the sieve specification is the decision that determines both performance and 20-year operating cost. The data in this guide should let you build a layered bed design that matches your specific syngas feed composition and product purity target. When you are ready to discuss your application in detail, reach out to the Aluminaworld technical team.

For 13X-HG, 5A, 4A, or matched activated alumina pre-bed kits for syngas service, contact us via:

• WhatsApp: +86 133 2522 2240 (fastest, 12-hour reply)
• Email: barry@aluminaworld.com
• Sample request: 25 kg layered-bed sample kit (13X-HG + activated alumina + 5A) with full CoA, 7-10 day lead time
• Bulk orders: 5 tonne MOQ, 20-30 day production, FOB/CIF/CFR from Qingdao Port (80 km from our factory)
• Site commissioning: Available for first-time installations in 60+ countries, including layered bed loading supervision and initial performance sampling

Aluminaworld has supplied molecular sieve to refinery and petrochemical hydrogen plants in 60+ countries for 15 years. Our 13X-HG is manufactured under ISO 9001 quality control with SGS on-site audits and full Alibaba Trade Assurance. Let us put our syngas PSA experience to work on your next project.